EXPERIMENTAL

CELL

RESEARCH

2%.?,113-124

(1992)

bril Assembly by Cornea1 Fibroblast n Gel Cultures: Small-Diameter sited in the Absence of Kerata KA~HLEE~J.I%ANE,*JOANNEP.BABIARZ,*JOHN *Department

of Pathology,

Robert Wood and CeGdar Biology,

M. FITCH,TTH

Johnson MedicaZ Tufts Uniuersity

School, School

Extracellular matrix assembly is a multistep process and the various steps in collagen fibrillogenesis are tbought to be influenced by a number of factors, including other noncollagenous matrix molecules. The synthesis and deposition of extracellular matrix by cornea1 fibroblasts grown within three-dimensional collagen gel cultures were examined to elucidate the factors important in the establishment of tissue-specific matrix architecture. Cornea1 fibroblasts in collagen gel cultures form layers and deposit small-diameter collagen fibrils (-25 nm) typical of the mature cornea1 stroma. The matrix syntbesized contains type VI collagen in a filamentous network and type I and type V collagen assembled as heterotypic fibrils. The amount of type V cobagen synthesized is relatively high and comparable to that seen in the cornea1 stroma. This matrix is deposited between cell layers in a manner reminiscent of the secondary cornea1 stroma, but is not deposited as densely or as organized as would be found in situ. No keratan sulfate proteoglycan, a proteoglycan found only in the cornea1 stroma was synthesized by the fibroblasts in the collagen gel cultures. The assembly and deposition of small-diameter fibrils with a collagen composition and structure identical to that seen in the cornea1 stroma in the absence of proteoglycans typical of the secondary cornea1 stroma imply that although proteoglycan-collagen interactions may function in the establishment of interfibrillar spacing and lamellar organization, collagen-collagen interactions are the major parameter in the regulation of fibril diameter. 0 1992

Academic

Press,

1~.

INTRODUCTION Connective tissue development requires the synthesis and deposition of collagen fihrils in a tissue-specific

’ To whom reprint requests should be addressed Pathology, Robert Wood Johnson Medical School, cataway, NJ 08854-5635. Fax: (908) 463-4847.

at Department of 675 Hoes Ln., Pis-

Piscataway,

of Medicine,

New Jersey, 08854.5635; Boston, Mussachusetts,

and TDepartment 02111.1800

o/Anatomy

manner. This assembly is a multistep lproeess and the various steps in fibrillogenesis are thought to be influenced by many factors, including ot nous matrix molecules. Determining the role of each factor is important in understanding the establishment of matrix architecture. The secondary stroma of the avian cornea consists of a monotony of small-diameter (~25 nm) fibrils arranged in orthogonal layers. The maintenance of constant fibril diameter ing of fibrils, and organization of these thogonal lamellae are important in the development of cornea1 transparency. The mature stroma contains heterotypic fihrils, consisting of types I and V collagen [I31, and has a relatively high percentage of type V collagen relative to that of other type I ~ollager~-containing tissues [4,5]. Type V collagen is important in .modulating fibril diameter [6, 71, and increasing the molar ratio of type V to type I collagen in an in d-0 se system leads to a decrease in fibril diameter. percentage of type V collagen comparable in the mature cornea1 stroma resulted in fibrils with larger diameters than those seen Z?Z situ [6], raising the possibility that other factors may be important in the regulation of cornea1 fibril diameterS In addition to collagen-collagen ~~teractio~~, procollagen processing can modulate fibri diameter, since the presence of type I cohagen propepti es at either the carboxy1 or the amino terminus ahers fibril assembly [all]. Cellular compartmentalization of the spatial and temporal events in collagen ~b~~~~oge~esi~also is important in the regulation of fibril formation, allowing the fibroblast to exert control over fibriI architecture by controlling the extracehular mixing and postdepositional processing of matrix components ]12 13]* Proteoglycan-collagen interactions also have been imphcated in the regulation of tibril diameter. The secondary corneai stroma contains two cl small9 fibril-asso ciated proteoglycans: one with atan sulfatelchondroitin sulfate side chains ) which has been shown to be decorin and a second with keratan sulfate side chains (KS PG) which has recently been termed

114

DOANE

lumican [14-171. Proteoglycans influence fibril assembly and tissue organization [18-231. In situ, changes in the relative amounts of specific proteoglycans can be correlated with alteration of fibril diameter [24, 251. Growth of cornea1 fibroblasts within three-dimensional collagen gels allows the study of collagen fibril formation in an in vitro system which can be easily manipulated, yet resembles the environment encountered by the fibroblast in uivo. Cells grown in three-dimensional collagen gel cultures appear morphologically similar to fibroblasts in viva, and within such an environment fibroblasts from different tissues orient in a manner characteristic of the tissue from which they were derived [26]. Fibroblasts grown in collagen gel cultures have been used as model systems for investigating extracellular matrix synthesis [27, 281 and since cornea1 fibroblasts grown in collagen gel cultures orient as they do in vivo, we reasoned that this system could be used to study cornea1 matrix synthesis and deposition. Here, we demonstrate that cornea1 fibroblasts synthesize and deposit a collagenous matrix containing heterotypic type I/V collagen fibrils of the correct cornea1 diameter. This matrix is deposited between cell layers in a manner reminiscent of the secondary cornea1 stroma. However, the cells do not synthesize detectable amounts of keratan sulfate proteoglycan, a proteoglycan found only in the cornea1 stroma. This indicates that keratan sulfate proteoglycans are not necessary for normal cornea1 fibril formation and implies that collagen-collagen interactions and not proteoglycan-collagen interactions are the major regulatory mechanism in the control of fibril diameter. MATERIALS

AND

METHODS

Cell culture. Cornea1 fibroblasts from 14-day embryonic white leghorn chickens were isolated and cultured within three-dimensional bovine type I collagen gels (Vitrogen 100, Collagen Corp.) as previously described [26]. Briefly, cells were isolated after bacterial collagenase treatment and were maintained for 7 days in minimal essential medium (GIBCO) with 0.15% sodium bicarbonate, 50 @g/ml gentamicin, and 2.5 pg/ml fungizone (CMEM) containing 10 or 20% fetal bovine serum (GIBCO) and 0 or 50 pg/ml ascorbate. Cells were trypsinized, counted, and then placed in wells containing 0.4 ml of 0.67 mg/ml neutralized Vitrogen in CMEM at a concentration of 105 cells per well. The collagen solution containing the cells was polymerized at 37’C such that the cells were suspended in the collagen gel. The edges of the collagen gels were released after 2-3 days in culture, allowing the gels to contract. In some experiments @-aminopropionitrile fumarate @APN) was present in the medium at 50 pg/ml to inhibit covalent cross-linking and thus make fibrils susceptible to unmasking by temperature manipulation [l, 291. Organ cukure. Corneas were excised from 13-14 day chicken embryos, rinsed in PBS with antibiotics (100 pg/ml gentamicin, 5 pg/ml fungizone), and cultured for 4 h in Dulbecco’s modified Eagle’s medium containing 0.5% fetal bovine serum, 50 pg/ml gentamicin, 2.5 @g/ml fungizone, 50 pg/ml ascorbate, and 0.15% sodium bicarbonate. The medium was replaced with medium containing either 10 pCi/ml [‘H]glucosamine or 50 pCi/ml “S as HZSOd (ICN Radiochemicals) and the cultures were incubated for 18 h. The corneas were rinsed in

ET

AL.

phosphate buffered saline (PBS), the central 2.mm dermal punch, and the proteoglycans

corneas extracted.

isolated

using

a

CoUogen c~~a~ys~s. Cornea1 fibroblasts were grown in conventional cell culture or in three-dimensional collagen gel cultures for 6 days, then labeled with 10 pCi/ml [3H]proline with or without 10 &i/ml [aH]glycine (ICN Radiochemicals) overnight. The gels/cell layer, the medium, and/or the gels plus the medium were analyzed. The gels/ cell layers were washed with PBS, pH 7.5, the wash was combined with the medium fraction, and the samples were dialyzed against 0.5 N acetic acid and then treated with 1.0 (gels) or 0.125 mg/ml pepsin overnight at 4’C. Collagens were lyophilized, resuspended in equal amounts of sample buffer, and electrophoresed on 5 or 6% SDS polyacrylamide gels [30]. Following electrophoresis, gels were fixed, impregnated with En3Hance, dried, and exposed to Kodak X-Omat film. Proteoglycan analysis. Cornea1 fibroblasts in collagen gels were grown in uitro for 6 days and then labeled with either 10 pCi/ml [‘HIglucosamine or 50 pCi/ml “S as H$Od (ICN Radiochemicals) for 18 h. Collagen gels and medium or central corneas from organ cultures were extracted in 4 &f guanidine HCl with 50 rnM sodium acetate, 10 rnM EDTA, 10 rnM benzamidine HCl, 10 m&f n-ethylmaleimide, and 0.1 Ji4 e-amino-N-caproic acid, pH 5.8, overnight at 4’C. Following centrifugation at 17,500g for 30 min at 4’C, the supernatant was reduced in volume with Aquacide (Calbiochem Corp.) and dialyzed against extraction buffer. Insoluble material was removed by centrifugation at 144,OOOg for 1.5 h at 4“C. Following extraction and solubilization of proteoglycans from collagen gels and medium, the supernatant was dialyzed into either 25 mA4 Tris-HCl, pH 7.8, or 20 mM sodium acetate, pH 5.0. Samples were digested with one of the following enzymes (Seikagaku America, Inc.): keratanase (0.2 units/ml), endo-b-galactosidase (0.02 units/ ml), chondroitinase ABC (0.1 units/ml), or Streptomyces hyaluronidase (3.3 units/ml). All enzyme digestions except the hyaluronidase were performed in 25 m&f Tris-HCl, pH 7.8. Keratanase digestion was performed at 37’C. Both endo-fl-galactosidase and chondroitinase digestion were performed at room temperature, as was the nonenzyme-digested control. Hyaluronidase digestion was performed in 20 m&f sodium acetate at 37’C. All material was digested for 24 h, and then dried using a Speed-Vat. Enzyme-digested or control proteoglycans were reconstituted using 4 i!4 guanidine HCl with 50 miU sodium acetate, pH 5.8. Samples were centrifuged in a Beckman Air-Fuge at 126,OOOg for 20 min and then equivalent cpms were loaded onto a Superose 6 column equilibrated in the same buffer. Fractions were collected and cpms were determined. Monoclonal antibodies. The following monoclonal antibodies were used I-BAl and I-DD4, anti-type I collagen antibodies [31, 321; VDH2, an anti-type V collagen antibody [33]; VI-EC6, an anti-type VI collagen antibody [34]; and IV-IAS, an anti-type IV collagen antibody [35,36]. These antibodies were specific for chick collagens and did not cross-react with bovine collagens. Zmmunofhorescence microscopy. Collagen deposition by cells grown in collagen gels was studied using indirect immunofluorescence with collagen type-specific monoclonal antibodies. Collagen gels were incubated in PBS containing 7.0% sucrose for 30 min and then embedded in OCT compound (Tissue-Tek; Miles Scientific), all at room temperature. The embedded tissue was frozen at -2O’C and cryostat sections were cut at 8 grn, picked up onto albumin coated microscope slides, and stored desiccated at -8O’C until used. Treated sections were exposed to 0.1 M acetic acid for 5 min at room temperature followed by fixation, while untreated sections were immediately fixed. Sections were rinsed briefly in PBS, free aldehydes were blocked using 0.5 mg/ml sodium borohydride in PBS for 15 min at room temperature, and nonspecific binding sites were blocked in 2% normal goat serum (NGS) for 2 h at room temperature. Antibodies were diluted to 25 pg/ml and the sections were incubated overnight at 4’C, washed, incubated with goat anti-mouse IgG labeled with FITC

CGRNEAL

COLLAGEN

FIBRIL

ASSEMBLY

or RKTC (Jackson Laboratories) diluted 1:40, washed, and mounted with FITC Guard (Testog, IL). All specimens were observed and photographed using a Zeiss Photomicroscope III equipped for epifluorescence, Kodak Tri-X film, and Diafine developer. Micrographs were prepared under identical exposure and printing conditions to permit comparisons. Transmission electron microscopy. Cornea1 fibroblasts in collagen gels were processed for transmission electron microscopy as previously described [6, 121. Briefly, collagen gels were fixed in 2.5% glutaraldehyde, 4% paraformaldehyde in 0.1 M sodium cacodylate, and 8 mM CaC&, pH 7.5, for 1 h on ice and then fixed in 1% osmium tetroxide in the same buffer for 1 h on ice. Following dehydration, en-bloc staining with ethanolic uranyl acetate, and infiltration, collagen gels were embedded in Epon. Sections were cut, stained with 2% aqueous many1 acetate and 1% PTA, pH 3.2, and viewed with either a Philips 420 or a JEOL 12OOEX transmission electron microscope. Ultracryomicrotomy. Cornea1 fibroblasts in collagen gels were also processed for ultracryomicrotomy. Geis were rinsed in PBS and fixed in 4 or 6% paraformaldehyde in 0.15 M sodium phosphate buffer, pH 7.4, containing 4.5% sucrose and 3% NaCl for 1 h at 4’C. Samples were infused with 1.8 M sucrose in 0.15 M sodium phosphate buffer, pH 7.4, with 0.05% azide over -1 h. A portion of the gel (-1 mm3) was positioned on specimen pins and frozen in liquid Nz. Thick sections (1 pm) were cut and stained with toluidine blue to check for orientation and cell density within ge!s. The blocks were trimmed to less than 0.5 mm’. The temperature of the cryochamber> block, and knife were then lowered to -90 to -1OO’C for ultrathin sectioning. Sections were cut at -70 nm, picked up on a sucrose loop, and transferred to carbon and formvar-coated nickel grids. Grids were stored specimen-side down on 2% gelatin in 0.15 M phosphate buffer with 0.05% azide at 4OC. Once enough grids were collected for immunolabeling, the gelatin was melted at 37OC. Grids were washed and quenched twice with 59 mM g!ycine in 0.15 M phosphate buffer with 0.05% azide at 37’C for 15 min. Tissues were blocked in 2% NGS in PBS at room temperature for 15 min. They were then incubated on a 25-~1 drop of primary antibody in a humid chamber at room temperature for 1 h. Grids were thoroughly washed in six changes of 50 mM glycine in 0.15 M phosphate buffer for 30 min. Grids were then incubated on a 25-~1 drop of secondary antibody-gold conjugate for I b at room temperature in humid chamber. Following washing with six changes of 50 mM glycine in 0.15 M phosphate buffer for 30 min, the grids were rinsed in two changes of l&O and stained with fresh uranyl acetate oxylate (pH 7-7.5) for 5 min. After two brief washes in HzO, the samples were infiltrated with 2% many1 acetate in methyl cellulose for 10 min on ice. Grids were looped and the methyl cellulose was drained using Whatman 54 hardened filter paper. The resulting films on grids had a gold-blue interference. Analysis of fibril diameter. To determine the diameter of fibrils deposited by cluck cornea1 fibroblasts in collagen gel cultures, fibrils localized adjacent to cells as well as immunolabeled ultrathin frozen sections and conventional plastic sections were analyzed. Fibril diameters were measured from calibrated photomicrographs. Diameter measurements were norma!ized using the internal 67 nm repeat of the fibrils. At least 50 fibrils from two different experiments were measured.

IN

3-D

GEL

CULTURES

-lETI

this matrix were very large and ha characteristic of fibrils formed in &Iization of bovine type 1 collagen A however, was observed at the surface o broblasts and between the layers of cells (Fig. I). This consisted of smaKdiameter fibri fiamentous material, and amorphous material, aracteristics which would be expected of a cornea1 ~~rob~~s~-derived matrix. Small-diameter fibrils were foun blasts. At a higher magnification, t fibrils found close to fibroblasts were uniform in diameter, and the 67 nm banding pattern was apparent (IZig. 1B). The fibrils were seen in both cross and longitudinal section. In addition, some regions where groups of fibrils were present with an orthogonal distribution were observed. This matrix associated with fibroblasts and composed of small-diameter fibrihs was less dense than that seen in developing 14-day yonic chick corneas. To analyze the deposition of gen fibrils by cells in collagen gel cultures, species- a type-specific monoclonal antibodies were used in ~~tras~r~~t~ra~ locahzation experiments Chicken type 1 collagen was localized to small-diameter fibrils close to cells (Figs. 2&2C)* The large-diameter fibrils, presumably .the bovine coIlagen gel, did not label with any of the anti-chicken collagen antibodies. ImmunoIabeling for type V collagen was not observed in the untreated gels (Fig. 2 against type IV were used as a egative codrol and all fibriks were nonreactive (F’ig~ 2 . hs addition cell-free gels demonstrated no reactivity for &i&en collagen types 1, IV, V, or VI (data not shown). Type VI collagen was localized to filaments which we present across fibrils and often close to ceils (Figs. 2 TQ determine whether the small chicken fibrils were heterotypic at the ultrastructural level? cornea1 fibroblasts in collagen gel cultures were treated with the latbyrogen @Al?N to inhibit cross-hnkmg of the newly synthesized collagen. Following swelhng of the tissue in the cold, fibril structure was ~art~a~~~ disrupted such that the helical epitope recognized by the monoclonal. antibody against type V collagen was unmasked. Indirect immunolabehng with anti-type V collagen antibody

type V collagen was distribute tures as a component of the s

RESULTS

Matrix deposition. To characterize the organization of cornea1 fibroblasts, fibril deposition, and matrix organization, cornea1 fibroblasts in collagen gel cultures were analyzed morphologically. Cornea1 fibroblasts grown m three-dimensional collagen gel cultures for 7 days formed several layers of cells within the artificial collagenous matrix- The collagen fibrils constituting

a condition where fibril integrity was maintame theie was little labeling for type V collagen (Fig. 3C). ouble labehng with anti-type V collagen (5 nm colloidal gold) and anti-type I coUagen (10 nm colloidal gold) antibodies demonstrated that both collagen types could be localized to the same fibril, indicatmg the presence of heterotypic fibrils (Figs0 4A-4C).

116

DOANE

Small-diameterfibrils. To characterize the fibrils deposited by cornea1 fibroblasts in hydrated collagen gel cultures with respect to fibril diameter, fibrils found between cell layers and adjacent to cells were measured from conventional plastic embedded transmission electron micrographs (Fig. 5A). This analysis revealed a narrow distribution of smalldiameter fib&. These fibrils had a mean diameter of 24.2 k 3.4 nm (MD) (Fig. 5B). When ultrathin cryosections were immunolabeled for chicken type I collagen, the small-diameter fibrils in these regions were positive. The diameters of these immunopositive fibrils were measured from micrographs and ranged from 17 to 27 nm. This indicates that corneal fibroblasts in three-dimensional collagen gels assemble and deposit fibrils with a diameter characteristic of the cornea1 stroma in si&. Collagen biosynthesis. Biosynthetic labeling and immunolocalization experiments were performed to characterize the fibrillar collagen types deposited into the matrix by cornea1 fibroblasts in three-dimensional collagen gel cultures. Cornea1 fibroblasts grown in collagen gels synthesized and assembIed types I and V collagen. Type I collagen was distributed throughout the matrix, while the presence of type V collagen in a similar distribution pattern could only be demonstrated after unmasking of the helical epitope with 0.5 M acetic acid (Figs. 6A-6D). A n antibody to type IV collagen, used as a negative control, was nonreactive (results not shown). In addition, bovine collagen gels containing no chicken fibroblasts were nonreactive for antibodies against chicken collagen types 1, V, and IV (data not shown). These data demonstrate the presence of both chicken collagen type I and type V within the collagen gel matrix. In addition, the unmasking required for labeling of type V collagen epitopes indicates that type I and type V collagen were present as heterotypic fibrils throughout the collagen gels. Biosynthetically labeled collagens were pepsintreated to convert procollagens to collagens followed by analysis using SDS-polyacrylamide gel electrophoresis (Figs. 6E-6F). The cornea1 fibroblasts grown on tissue culture plastic secreted type V collagen in an amount comparable to that seen in the cornea1 stroma. In addition, the cell-populated collagen gels and medium were analyzed. The gel matrix contained between 60 and 80% of the total incorporated radioactivity, indicating that

ET

AL.

fibrils were being incorporated into the gel matrix. Fluorograms indicated that collagens types I and V were synthesized by cornea1 fibroblasts in three-dimensional collagen gel cultures. Type V collagen represented a relatively high and constant proportion of the newly synthesized fibrillar collagen with the type V collagen present as a chains al(V) and a2(V) present in a 25 ratio. Proteoglycan biosynthesis. The proteoglycan species produced by cornea1 fibroblasts grown in hydrated collagen gel cultures were determined after biosynthetic labeling, enzyme digestion, and gel filtration chromatography. Proteoglycans extracted from corneas labeled in organ culture or from cornea1 fibroblast-populated collagen gels were separated by gel filtration chromatography on Superose 6. The 35S-labeled proteoglycans from cornea1 organ cultures chromatographed as a single included peak with a Kaveof 0.21 (Fig. 7A). In contrast, the majority of “S-labeled proteoglycans extracted from collagen gel cultures chromatographed as a highermolecular-weight peak, eluting in the void volume (Fig. 7B). Proteoglycans biosynthetically labeled with ‘% were extracted from corneas and collagen gel cultures and were analyzed after hydrolysis with chondroitinase ABC, keratanase, and endo-/Y-galactosidase. The chromatograph obtained after keratanase or endo-fl-galactosidase digestion of proteoglycans from collagen gel cultures (Fig. 7H) was virtually the same as the profile of the undigested material (Fig. 7B), indicating that little of the sulfated material was keratan sulfate proteoglycan. In contrast, keratanase or endo-fl-galactosidase digestion of cornea1 proteoglycan extracts hydrolyzed a significant amount of the labeled proteogly~ans, shifting the digested material to the total column volume (Fig. 7G). Chondroitinase ABC treatment of proteoglycans from both cornea and collagen gel cultures hydrolyzed a significant percentage of the labeled material in each case, causing a shift to a lower-molecular-weight species which chromatographed at the total volume (Fig. 7D and 7E). Virtually all of the 35S-labeled proteoglycans from collagen gels were hydrolyzed, indicating that dermatan sulfate proteoglycan was the only sulfated macromolecule synthesized. In organ cultures, however, only a portion of the cornea1 proteogly~ans were affected. The amount was consistent with the keratanase digestions

FIG. 1. Cornea1 fibroblasts grown in collagen gel cultures formed layers and deposited a fibrillar extracellular matrix. (A) In this transmission electron micrograph, cornea1 fibroblast formed layers within the exogenous collagen gel (curved arrows). Small-diameter fibrils were seen adjacent to and between these cells in both cross and longitu~nal section (open arrows). Small groups of fibrils were occasionally seen with an orthogonal arrangement. Filamentous material (*) also was deposited between the fibroblasts. In (B) a cornea1 fibroblast within the exogenous bovine collagen matrix (curved arrows) is seen at higher magnification. Numerous small-diameter fibrils (open arrows) were adjacent to the fibroblast. Note t.he much larger diameter of the bovine type I collagen Gbrils which constitute the three-dimensional collagen gel (curved arrow). l3ar, 1.0 pm.

CQRNEAL

COLLAGEN

FIBRIL

ASSEMBLY

IN

3-D

GEL

CULTURES

117

118

DOANE

and indicates that this peak contains both dermatan sulfate proteoglycan and keratan sulfate proteoglycans. Proteoglycans in collagen gel cultures also were biosynthetically labeled with [3H]glucosamine to determine whether nonsulfated or poorly sulfated keratan sulfate proteoglycan or hyaluronic acid were present. Proteoglycans were separated by gel filtration chromatography on Superose 6. The extracted proteoglycans chromatographed as a major peak eluting in the void volume and a minor peak eluting with a Kave of 0.43 (Fig. 7C). Chondroitinase ABC hydrolyzed most of the [3HJglucosamine-labeled material extracted from collagen gel cultures, causing a shift to the total volume (Fig. 7F). The remainder of the major peak chromatographed in the column void. Hyaluronidase digestion caused a shift of -15% of the material in the major peak to a lowermolecular-weight species (results not shown). This amount is consistent with the chondroitinase ABC insensitive material. We were unable to identify the minor peak. This indicates that most of the material was dermatan sulfate proteoglycan (Fig. 7F) and that cornea1 fibroblasts in collagen gels were also synthesizing hyaluranic acid (results not shown). Keratanase and endofi-galactosidase, which recognizes nonsulfated keratan sulfate, were also used to digest [3H]glucosaminelabeled proteoglycans. The profiles obtained with these two enzymes were virtually identical. Both enzymes hydrolyzed a small amount of ‘H-labeled material which eluted at the total column volume (Fig. 71). This indicates that very little, if any, of total proteoglycans from cornea1 fibroblasts in collagen gels was under- or nonsulfated keratan sulfate proteoglycan. DISCUSSION

The synthesis and deposition of extracellular matrix by cornea1 fibroblasts grown within three-dimensional collagen gel cultures was examined to elucidate the factors important in the establishment of tissue-specific matrix architecture. It has been previously shown that cornea1 fibroblasts grown under these conditions maintain their tissue-specific orientation, forming orthogonal cell layers [26]. Here, we demonstrate that cornea1 fibroblasts establish a tissue-specific configuration, synthesize collagen types I and V in amounts similar to that seen in situ, as well as collagen type VI, assemble collagen fibrils typical of the secondary cornea1 stroma, and

FIG. 2. Immunocytochemical localization of collagens deposited by V, and VI were localized in ultrathin frozen sections using species- and labeled with colloidal gold. Immunolabeling with antibodies to chicken contained this collagen type (A-C). The larger fibrils of the bovine type magnification, there was a periodicity to the labeling with these two collagen was unreactive in normal, nontreated tissues (D). Antibody to fibrils (G). An antibody to type IV collagen, used as a negative control, staining. Bar, (A) 500 nm; (B-F) 250 nm; (G) 100 nm.

ET

AL.

incorporate these into a collagenous matrix. At the same time, no keratan sulfate proteoglycan is synthesized and the dermatan sulfate proteoglycan characteristic of the cornea in situ is altered. These data imply that for cornea1 fibril assembly, collagen stoichiometry and cellular topography are essential while cornea1 proteoglycans are not necessary. The high proportion of type V collagen present in the cornea, the presence of this collagen throughout the secondary stroma in heterotypic fibrils, and the presence of type V collagen in the typical quarter-stagger array [l31, argue that type V collagen is important in regulating cornea1 fibril diameter. Type V collagen may regulate fibril diameter due to its longer helix, which may interfere with lateral fibrillar growth [37, 381. In addition, type V collagen has an amino terminal globular domain which is retained after fibril assembly, and this domain is necessary for the full-diameter-regulatory effect seen in self-assembly assays. The location of this domain on the fibril surface may sterically inhibit lateral growth of the fibril[39]. Using an in uitro self-assembly system, we have previously shown that collagen-collagen interactions are important in the control of fibril diameter. In this assay, increasing the molar ratio of type V collagen to type I collagen produced fibrils of decreasing diameter [6]; however, the same concentration of type V collagen found in the cornea produced fibrils with a larger diameter than seen in situ. This suggested that either this artificial system does not accurately reflect the situation in situ (i.e., absence of procollagen processing, etc.) or that other mechanisms are involved in the regulation of fibril assembly. The three-dimensional collagen gel culture system employed in this study allowed the study of matrix assembly in uitro. On planar substrates cornea1 fibroblasts do not deposit an extensive fibrillar matrix as do these same fibroblasts cultured in three-dimensional collagen gel cultures and do not efficiently process procollagen, much of which is lost to the medium [5]. However, in the three-dimensional collagen gel cultures the cornea1 fibroblasts incorporate between 60 and 80% of synthesized collagen into the gel matrix. ln addition, within this culture system, matrix assembly presumably occurs as a multistep process as seen irz viuo. Our results demonstrate the absence of keratan sulfate proteoglycans normally seen in situ when the corneal fibroblasts are grown in the collagen gel cultures.

cornea1 fibroblasts in hydrated collagen gel cultures. Collagens types I, type-specific monoclonal antibodies followed by a secondary antibody type I collagen demonstrated that small-diameter fibrils close to cells I collagen matrix were not labeled with this antibody (A). At a higher monoclonal antibodies to type I collagen (B, C). Antibody to type V type VI collagen was localized to filaments close to cells (F) and across did not label small-diameter fibrils (E) and exhibited no nonspecific

CORNEAL

COLLAGEN

FIBRIL

ASSEMBLY

IN

3-D

GEL

CULTURE§

120

DOANE

ET

AL.

FIG. 3. Monoclonal antibody to type V collagen stained small-diameter fib& after disruption of fibril structure. Cornea1 fibroblasts were grown in collagen gel cultures in the presence of OAPN to inhibit covalent cross-linking of newly synthesized chicken collagens. The collagen gel cultures were treated at 4’C, which partially disrupted collagen fibril structure, followed by fixation and preparation of cryosections. When ultrathin frozen sections were labeled with antibody to type V collagen followed by a secondary antibody, uniform labeling of small-diameter fibrils present between cells was demonstrated (A). A comparison between cold-treated lathyritic collagen gels (B) and untreated lathyritic gels (C) iliustrates little immunostaining for type V collagen unless the helicai epitope recognized was exposed by partial disruption of fibrils. Bar, (A) 250 nm; (B, C) 100 nm.

However, fibrils are formed with a diameter characteristic of that seen in the cornea1 stroma. This indicates that KS PG-collagen interactions are not involved in the regulation of this stage of matrix assembly. In addition, the small dermatan sulfate PG (decorin) characteristic of the cornea1 stroma is abnormally large in the culture system. This indicates that the tissue-specific form is not required for diameter regulation, but can not

rule out an influence by the abnormal form produced. However, we have used @-D-xyloside to inhibit DS PG synthesis in situ [40]. This study demonstrated that DS PG is not involved in the regulation of fibril diameter, but does have a role in the development of matrix architecture, specifically the establishment of fibril packing and maintenance of lamellar integrity [40]. Coupled with other studies [lS, 411 this indicates that DS PG is

C~RN~AL

COLLAGEN

FIBRIL

AS§EM~i,Y

IlV

3-D

GEL

CAPTURES

121

proteoglycan from tendon has a high affinity for type 1 collagen and its s~e~~~c~ty of bin ing &es in the protege core [23]. It has been suggested that small proteoglycans bound to the fibril surface are mediators of fibril growth- Lateral fusion of fibrils, K~s~~~~~~in larger-d&-

FIG* 4. Collagen types I and V deposited by cornea1 fibrobiasts in three-dimensionai collagen gel cultures interact to form beterotypic fibrils. Monoclonal antibodies to type I collagen direct labeled with 1.0 nm colloidal gold and antibody to type V collagen direct labeled with 5 nm colloidal gold were both localized to the same fibril on ultrathin frozen sections of lathyritic, cold-treated collagen gel cultures (A-C). The short arrows indicate 5.nm gold label while the long arrows indicate lo-nm colloidal gold particles. &IF, 100 nm.

not required for normal cornea1 fibril assembly and diameter regulation. In a variety of systems proteoglycan-collagen interactions have been suggested as being ~rn~orta~t in the re~~~t~o~ of fibril assembly and tissue organization [B-23]. The secondary cornea1 stroma contains dermatan sulfate and keratan sulfate proteoglyeans which interact with specific regions on the collagen fibril [42]. It has been reported that in some tissues a re~at~ons~~~ exists between the relative content of specific proteoglycans and fibril diameter [24,25,43]. The effects of proteoglycans in collagen fibril formation also have been studied in z&o using proteoglycans from cornea or sclera [l$] and were shown to affect the kinetics of fibril formation differently, but no differences m the fibril diameter were observed. However, other studies have reported a modulation in fibril diamet,er in different tissues ]21]. The small dermatan sulfate

EIG. 5. Cornea1 fibroblasts in three-dimensionai collagen gel cultures deposited collagen fibrils with ~~~rneters typkal of the carneal stroma kz sk%. This high magnification transmission electron micrograph (A) illustrates the banding pattern of small-diameter fibrils. A histogram of fibril diameters measured from micrographs from two different experiments demonstrates that the small-diameter fibrils deposited by the corneai fibroblasts range primarily from l&-30 nm, with a meen of 24.18 k 3.36 nm (B), This suggests that cornea1 fibroblasts in collagen gel cultures deposit a fibrillar collagenous matrix characteristic of the secondary corneaI stroma. Bar, 1QQ nm.

122

DOANE

ET

AL.

FIG. 6. Cornea1 fibroblasts in three-dimensional collagen gel cultures synthesize collagen types 1 and V and deposit heterotypic collagen type I/V fib& into the gel matrix. Collagen types I and V were localized using immunofluorescence microscopy with collagen type- and species-specific monoclonal antibodies. Chicken type I collagen was localized throughout the collagen gel under conditions that either maintain (A) or disrupt flbril structure with acid treatment (C). However, the helical epitope which the type V collagen monoclona1 antibody recognizes was masked in untreated collagen gels (B). Acid treatment partiaily disrupts fibril structure, and type V collagen is seen throughout the collagen gel (D), with a similar distribution to that seen for type I collagen. Bar, 10 ,um. Collagens synthesized by cornea1 ~broblasts in cell labeled, run on SDS-polyacrylamide gels, and visualized using fluorograculture (E) or 3-D collagen gel cultures (E, F) were biosynthetically phy. Type V cohagen constituted a percentage of the fibrillar collagens comparable to that seen &r situ in both cell culture and collagen gel cultures. Chick embryo cornea1 fibroblasts in cell culture, cell layer (E, lane l), and medium (E, lane 2) or 3-D collagen gel cultures, gels, and medium (E, lane 3). Cornea1 fibroblasts in 3-D collagen gel cultures (F). Medium (pepsin soluble) from gel culture with 12.5 X 103 cells (F, lane 1). Medium (pepsin soluble) from gel culture with 25 X 103 cells (I?, lane 2). Medium (pepsin insoluble) from ge1 culture with 25 X 10s cells (F, lane 3). Medium and gel from culture with 25 X 1Oa cells (F, lane 4).

meter fibrils, may occur when proteoglycan sheaths are removed from the fibril surface [25]. Our data, however, suggest that such interactions are not necessary for the assembIy and maintenance of small-diameter cornea1 fibrils. Procohagen processing is an important parameter in the regulation of fibril diameter. Propeptides are generally transient elements whose rate of removal depends on collagen type and monitions of deposition, Unprocessed propeptides significantly alter fibril assembly [8, 9]. The amino-propeptides of types I and III collagen

have been found in association with thin flbrils in developing skin [IO, 111 and a variety of other tissues. However, the collagen type I propeptides are not maint~ned on fibrils within the cornea1 stroma [44]. In u&o, fibril diameter also has been shown to be affected by the rate and order of procollagen cleavage 18,451. We have not addressed the issue of procollagen processing in our three-dimensional collagen gel system; however, we believe this mechanism to be a transient regulatory mechanism in developing systems where alterations in fibril diameter occur. For example, in developing skin, initial

Fraction Nurrber

Fraction

Number

Fracwm

r

FIG. 7. Cornea1 fibroblasts grown in three-dimensional collagen gel cultures synthesize proteoglycans uncharacteristic of the normal cornea1 stroma. Proteoglycans from chick corneas grown in organ culture (A, D, G) and from cornea1 fibroblasts grown in collagen gel cultures (B, C, E, F, H, I) were biosynthetically labeled with either “S (A, B, D, E, G, H) or [‘H]glucosamine (C, F, I). The extracted proteoglycans weye ~hromatographed on Superose 6 after no enzymatic treatment (A-C), chondroitinase ABC treatment (D-F), keratanase (G-1, s&d ibx), or endo-~-g~actos~dase (G-I, dashed like). These data indicate cornea1 fibroblasts in collagen gel cultures synthesize a larger dermatan sulfate proteoglycan than that seen irz situ. In addition, little, if any, keratan sulfate proteoglycan is synthesized by t&e fibroblasts in collagen gels. The Pharmacia Superose 6 coh~mn has a V0 of 8.0 ml and a Vt of 22.0 ml. Samples were cbromatographed in 4.Q M guanidine HC$50 mM sodium acet.ate$ pH 5.7, and BSA (67 kDa), aldolase (158 kDa), catalase (232 kDa), and thyroglobulin (669 kDa) were run for reference and had a KaVe of 0.5, 0.35, 0.28, and O.Q5, respectively

deposition of small-diameter fibrils which retain pro” peptides is folIowed by cleavage of these domains. This leads to a lateral tision of fibrils, resulting in an increase in diameter. In the corn& stroma, fibrils are a consistent small diameter, and thus a different, permanent mechanism must ~nn~~ion. We suggest that propeptides are not retained on type I collagen molecules in LGUO in the secondary cornea1 stroma since the presence of type V collagen, w&b retention of its amino terminal globular domain, may function in a way similar to retention of propeptides t,o permanently regulate deposition of small-diameter fibrils. Even though cornea1 fibroblasts grown in hydrated

collagen gel cultures assemble ad deposit collagen fiW brils wi& well-regulated diameters ~~~sis~en~ with those seen in S&L, these Gbrils are not deposited in wellorganized orthogonal arrays at. a de~§~t~ typical of the secondary cornea1 stroma. Thus, a~t~~~g~ h&b& carneal ~ro~eogly~ans are not necessary for control of fibril diameter, it is possible that they f~~~~io~ in the assembly of higher order stru&ures such as bundles with constant interfibrillar spacing and IameHae [40]~ Mthough cornea1 fibroblasts in coILagen gel cultures produced a larger-molecular-weight dermatan sulfate proteoglycan and little, if any? keratan sulfate ~r~~~o~ly~a~, it is possible that the core protein of either one or both proteogly-

124

DOANE

ET

cans was present, and could have influenced fibril diameters. Heterotypic type I/V collagen fibrils 25 nm in diameter, characteristic of the cornea1 stroma in situ, have previously never been demonstrated in an in vitro system. The current study indicates that the temporal and spatial mixing of the correct ratio of fibrillar collagens into heterotypic fibrils, as controlled by the cornea1 fibroblast, is important in the control of cornea1 fibril diameter. Since these fibrils were deposited in the presence of a dermatan sulfate proteoglycan uncharacteristic of the cornea1 stroma and no keratan sulfate proteoglycan, we conclude that these matrix components are not necessary for the assembly and deposition of smalldiameter heterotypic fibrils by cornea1 fibroblasts in three-dimensional collagen gel cultures. This implies that collagen-collagen interactions predominate in determining collagen fibril structure. We thank Rita Hahn and Susan Sommers for expert technical assistance and Dr. Robert Trelstad for critically reading the manuscript. This work was supported by NIH Grants EY 05129 (D.E.B.) and EY 05191 (T.F.L.) and National Research Service Award EY 06227 (K.J.D.).

14.

Blochberger, T. C., Vergnes, J.-P., (1991) J. Biol. Chem. 267,347-352.

15.

Funderburgh, J. L., Caterson, Biol, 116,267-277. Midura, R. J., and Hascall,

16.

Birk, D. E., Fitch, (1988) J. Cell B&l.

2.

Fitch, J. M., Birk, D. E., Mentzer, A., Hasty, K. A., Mainardi, C., and Linsenmayer, T. F. (1988) Invest. Ophthalmol. Vis. Sci. 29, 11251136. Linsenmayer, T. F., Fitch, J. M., and Birk, D. E. (1990) Ann. N. Y. Acad. Sci. 580, 143-160.

3.

J. M., Babiarz, 106, 999-1008.

J. P., and Linsenmayer,

4.

Davison, Rex 29,

5.

McLaughlin, Cell Sci. 94,

6.

Birk, D. E., Fitch, J. M., Babiarz, J. P., Doane, mayer, T. F. (1990) J. CelZ Sci. 95, 649-657.

7.

Adachi,

8.

Miyahara, R. L., and Mould, A. C. H., and Fleischmajer, Wiestner, Acad. Sci.

9. 10. 11. 12.

P. F., Hong, 97-107.

B. S., and Cannon,

J. S., Linsenmayer, 371-379.

T. F.

D. F. (1979)

T. F., and Birk,

Exp.

17. 18. 19.

D. E., and

Trelstad,

R. L. (1984)

J. Cell Biol.

99,

Received Revised

March version

18, 1992 received May

4, 1992

R. L, (1991) E. D., Ed.)

D. E., and Lande,

J. BioZ. Chem.

Biophys.

Actu

251,6513-6521.

M. A. (1981)

Biochim.

Vogel,

K. G., Paulsson,

M., and Heinegard,

D. (1984)

Biochem.

J.

L. W. (1987)

21.

Vogel,

Coil. ReZat. Res. 7, 105-

22. 23. 24.

Ruoslahti,

25. 26. 27.

K. G., and Trotter,

J. A. (1987)

Biochem.

114. E. (1988)

Annu.

Rev. Cell Biol.

4, 229-255.

Brown, D. C., and Vogel, K. G. (1990) Matrix 9,468-478. Borcherding, M. S., Blacik, L. J., Sittig, J. W., Bizzell, M. B., Breen, M., and Weinstein, H. G. (1975) Exp. EyeRes. 21,59-70. Scott, J. E. (1984) Biochem. J. 195, 229-233. Deane, K. J., and Birk, D. E. (1991) Exp. CeZZ Res. 195,432-442. Mochitate,

K., Pawelek,

P., and Grinnell,

F. (1991)

Exp. Cell Res.

193,198-207. 28.

Coustry, F., Gillery, P., Maquart, FEBS Lett. 262,339-341.

29.

Fitch, J. M., Gross, J., Mayne, senmayer, T. F. (1984) Proc. 2795.

30. 31.

Laemmli, U. K. (1970) Nature 227, 680-685. Linsenmayer, T. F., Gibney, E., and Fitch, J. M. Biol. 103,1587-1593.

32.

Linsenmayer, T. F., Hendrix, M. J., and Little, C. D. (1979) Proc. Natl. Acad. Sci. USA 76,3703-3707. Linsenmayer, T. F., Fitch, J. M., Schmid, T. M., Sanderson, R. D., and Mayne, R. (1983) J CeZZ BioZ. 96,124-132. Linsenmayer, T. F., Mentzer, A., Irwin, M. H., Waldrep, N. K., and Mayne, R. (1986) Exp. Cell Res. 165,518-529.

33.

F.-X.,

and Borel,

J.-P.

(1990)

R., Johnson Wint, B., and LinNatl. Acad. Sci. USA 81, 2791-

(1986)

J. Cell

Fitch, J. M., Gibney, E., Sanderson, senmayer, T. F. (1982) J. Cell Biol.

36.

Mayne, R., Sanderson, R. D., Wiedemann, H., Fitch, J. M., and Linsenmayer, T. F. (1983) J. Biol. Chem. 258,5794-5797. Bachinger, H. P., Doege, K. J., Petschek, J. P., Fessler, L. I., and Fessler, J. H. (1982) J. BioZ. Chem. 257, 14590-14592. Silver, F. H., and Birk, D. E. (1984) Int. J. BioZ. Macromol. 6,

40. 41. 42. 43.

2024-

in Cell Biol2nd ed., pp.

R. D., Mayne, 95, 641-647.

R., and Lin-

125-132. 39.

44.

Birk, D. E., Silver, F. H., and Trelstad, ogy of the Extracellular Matrix (Hay, 221-254, Plenum, New York.

G. W. (1976)

Birk,

35.

2033. 13.

Hart,

J.

Fleischmajer, R., Olsen, B. R., Timpl, R., Perlish, J. S., andLovelace, 0. (1983) Proc. NatZ. Acad. Sci. USA SO, 3354-3358. Birk,

264,

Vogel, K. G., Koob, T. J., and Fisher, Biophys. Res. Commun 148,658-663.

38.

Res.

Chem.

34.

K. J., and LinsenTissue

J. Biol.

Deu.

20.

M., Hayashi, K., Berger, J., Tanzawa,F. K.,Trelstad, Prockop, D. J. (1984) J. Biol. Chem. 259,9891-9898. P., Hulmes, D. J., Holmes, D. F., Cummings, C., Sear, Chapman, J. A. (1990) J. MoZ. BioZ. 211,581-594. R., Timpl, R., Tuderman, L., Raisher, L., M., Perlish, J. S., and Graves, P. N. (1981) Proc. NutZ. USA 78, 7360-7364.

Connect.

V. C. (1989)

G. W. (1986)

223,587-597.

37.

T. (1986)

B., and Conrad,

J. R.

6'70,362-369.

14,257-

E., and Hayashi,

J., and Hassell,

Eye

D. E. (1989)

266.

Hemple,

1423-1430.

REFERENCES 1.

AL.

45.

Broek, D. L., Madri, J., Eikenberiy, E. F., and Brodsky, B. (1985) J. Biol. Chem. 260, 555-562. Hahn, R. A., and Birk, D. E. (1992) DeueZopment, 115, in press. Bansal, M. K., Ross, A. S. A., andBard, J. B. L. (1989) Dev. Biol. 133, 185-195. Scott, J. E., and Haigh, M. (1988) Biochem. J. 253, 607-610. Merrilees, M. J., Tiang, K. M., and Scott, L. (1987) Connect. Tissue. Res. 16, 237-257. Mellor, S. J., Atkins, G. L., and Hulmes, D. J. S. (1991) Biochem. J. 276, 777-784. Miyahara, M., Njieha, F. K., and Prockop, D. J. (1982) J Biol. Chem. 257,8442-8448.