Proc. Nati. Acad. Sci. USA Vol. 83, pp. 5126-5130, July 1986 Cell Biology In vitro regulation of cartilage matrix assembly by a Mr 54,000 -binding protein (type H collagen fibril assembly/proteoglycans) SRINIVASAN CHANDRASEKHAR*, GORDON W. LAURIE, FRANCES B. CANNON, GEORGE R. MARTIN, AND HYNDA K. KLEINMAN Laboratory of Developmental Biology and Anomalies, National Institute of Dental Research, National Institutes of Health, Bethesda, MD 20892 Communicated by Victor A. McKusick, March 18, 1986

ABSTRACT In cartilage, type II collagen is present as is not noticeably changed in the cartilage of a chicken mutant thin, short, randomly oriented fibrils. In vitro, however, type that lacks proteoglycan (34). Such observations indicate that II collagen forms fibrils of large diameter, indicating that proteoglycans may not be the most important regulatory additional factors may be involved in the regulation of collagen molecule in determining fibril size. fibril formation. We have examined extracts of a cartilage- In this study, we have identified a collagen-binding protein producing tumor for the presence ofcollagen-binding proteins. of Mr 54,000 in extracts of a well-characterized chondrosar- In addition to fibronectin and link protein, a Mr 54,000 protein coma that produces cartilage macromolecules. We call this was found to bind to collagen fibrils as well as to native and Mr 54,000 protein CBP. Immunological studies indicate that denatured type II collagen. Immunological studies using anti- CBP is also a constituent of normal cartilage but not of other body against the protein indicate that it is a cartilage matrix tissues. CBP binds to type II collagen and, in combination protein, not present in bone or in several other tissues. In vitro with cartilage proteoglycan, limits the growth of the type II studies show that the Mt 54,000 protein in combination with collagen fibrils. This protein may regulate fibril size in cartilage proteoglycan decreases the rate of type II fibril cartilage. formation and causes the fibrils to be of small diameter (24 ± 8 nm). These studies indicate that complexes between collagen MATERIALS AND METHODS and proteoglycans mediated by this protein may regulate the Isolation and Purification of CBP. The Swarm chondro- assembly of cartilage matrix. sarcoma (35) was used as the source of cartilagenous pro- teins. The tumor was grown in rats and harvested about 4 Cartilage has a distinctive histology, containing chondro- weeks after inoculation. All subsequent procedures were cytes surrounded by a rather homogeneous matrix. The performed at 40C unless otherwise indicated. Freshly har- matrix is composed of small collagen fibrils (1-3) widely vested tissue (200 g/liter) was homogenized in a solution separated by large aggregate structures of proteoglycan, containing 0.05 M Tris'HCl (pH 7.2), 20% NaCl, 0.01 M hyaluronic acid, and link protein (4-10). This arrangement of EDTA, 0.1 M 6-aminohexanoic acid, and 5 mM benzami- collagen fibrils and space-filling proteoglycan aggregates dine HCl to remove serum contaminants and was centrifuged generates a cushioning matrix that resists compression. It is at 25,000 x g for 30 min. The residue was reextracted with 500 not known whether or how collagen fibrils and proteoglycan ml of 2 M urea/0.05 M Tris-HCl, pH 7.2, containing the aggregates are linked in the matrix. protease inhibitors used above. This latter extract was used Most (>90%) of the collagen in cartilage matrix is type II as the source of collagen-binding proteins. collagen, although certain other such as types IX Portions of the urea extract were chromatographed on a and X and the la, 2a, and 3a collagens are also present column (25 x 5 cm) of DEAE- equilibrated with the (11-13). Type II collagen is secreted from chondrocytes as a same solvent. The unbound fraction, containing CBP, was soluble precursor, known as type II procollagen, that is dialyzed against 2 M urea/0.,15 M NaCl/5 mM phosphate, pH rapidly converted to type II collagen by specific proteases. 6.8, and was chromatographed on a column (25 x 5 cm) of Under physiological conditions, collagen is insoluble and hydroxyapatite (Ultrogel; LKB) equilibrated in the same assembles into fibrils (14). Type I collagen has the capacity buffer. Bound materials were eluted with a linear gradient to to self-assemble into fibrils in vitro with the same packing of 0.3 M phosphate in the same buffer. molecules in the fibrils as that observed in vivo (15-24). This Type II Collagen. Type II collagen was isolated from the is also true with type II collagen, but the fibrils obtained are chondrosarcoma grown in lathyritic rats (35). This material very large in diameter, in contrast to the small fibrils contained a small amount (usually <2%) of type I collagen, observed in hyaline cartilage (25, 26). These observations which has been shown to originate from the capsule. In some suggest that information for fibril formation is inherent in the studies, native or denatured type II collagen was coupled to molecule and/or that other factors exist that regulate the size CNBr-activated Sepharose (Pharmacia). of Collagen Fibrillogenesis. Type II collagen (1 mg/ml in 0.5-M the collagen fibrils in cartilage. acetic acid) was dialyzed against 0.01 M phosphate buffer Observations on type I collagen fibril formation indicate (pH 6.8) at 40C overnight to induce fibrils to form. Fibril- that several noncollagenous glycoproteins and proteoglycans logenesis was also initiated by incubation at 370C, after a influence fibril assembly (27-33). The large chondroitin solution of type II collagen (in 0.5 M acetic acid) was mixed sulfate proteoglycan ofcartilage and its isolated glycosamino- at 40C with a concentrated buffer solution such that the final glycan side chains cause some alterations in the kinetics of concentration of collagen was 0.6 mg/ml in 0.135 M type I fibril formation but do not reduce the diameter of the NaCl/0.03 M sodium phosphate/0.02 M N-tris(hydroxy- fibrils (33). Further, the diameter ofthe type II collagen fibrils *To whom correspondence should be addressed at present address: The publication costs of this article were defrayed in part by page charge Lilly Research Laboratory, Building 98, Room 4331, Division of payment. This article must therefore be hereby marked "advertisement" Immunology and Research, 307 East McCarty in accordance with 18 U.S.C. §1734 solely to indicate this fact. Street, Indianapolis, IN 46285. 5126 Downloaded by guest on September 29, 2021 Cell Biology: Chandrasekhar et al. Proc. NatL. Acad. Sci. USA 83 (1986) 5127 methyl)methyl-2-aminoethanesulfonic acid, pH 7.4. The ef- fects of CBP and cartilage proteoglycan on collagen fibril- logenesis were examined at 50 ,ug and 100 Ag/ml, respec- 200- - tively, by measuring changes in the optical density of the sample in a Gilford spectrophotometer at 314 nm (15, 25, 30). 116- s-l for 6 hr - Aliquots taken from these samples after incubation 93- at at 350C were diluted 1:30 with distilled water; 10 A.l of the diluted samples were placed on a Formvar-covered grid, stained with 1% uranyl acetate for 5 min, and examined in a 68 - 1 -mm JEOL 100C electron microscope. For quantitation of fibril diameter, electron microscopic -54 negatives of incubations of type II collagen alone, type II 'U collagen plus chondroitin sulfate proteoglycan, type II 42- eU collagen plus CBP, or type II collagen plus chondroitin sulfate proteoglycan and CBP were projected onto a ZIDAS 1 2 3 4 5 (Zeiss) digitizing tablet at a final magnification of200,000. On the tablet, a grid was drawn with lines separated by 4 cm. FIG. 1. Binding of CBP to collagen fibrils. The 2 M urea extract Fibrils that fell under the intersection of two lines were then was dialyzed against PBS and centrifuged, and the supernatant to determine diameter. fraction (0.5 mg of protein in 0.5 ml) was incubated with or without measured collagen fibrils (1.0 mg in 0.5 ml) at 350C for 15 min. After incubation, Preparation of Antibodies to CBP. The 2 M urea extract of the solutions were centrifuged and the pellets were examined by chondrosarcoma was dialyzed against phosphate-buffered NaDodSO4/7.5% PAGE in the presence ofa reducing agent. Lane 1: saline (PBS, 0.02 M sodium phosphate, pH 7.2/0.15 M NaCl) molecular weight standards (M, x 10-3 at left). Lane 2: 2 M urea and 'centrifuged to remove insoluble material. The superna- extract. Lane 3: the material precipitated when the 2 M urea extract tant fluid was mixed with type 1Lcollagenfibrils (prepared as was incubated alone. -Lane 4: the material precipitated when type II described above) and incubated at 370C for 1 hr. The fibrils were incubated alone. Lane 5: the material precipitated when precipitate was collected by centrifugation at 25,000 x g for type II fibrils and 2 M urea extract were incubated together. Position 15 min, washed with PBS, solubilized in 1% NaDodSO4, and of CBP (M, 54,000) is indicated at right. subjected to electrophoresis. Subsequently, the portion ofgel containing CBP (100 Izg in 1.5 ml) was mixed with complete coupled to Sepharose. Bound material was eluted with 6 M Freund's adjuvant and injected into rabbits. At least three urea and was found to consist of three major components booster injections of the same material in incomplete adju- (Fig. 2, lane 3). These were (i) the Mr 54,000 protein observed vant were administered at 2-week intervals. above; (ii) a Mr 220,000 protein, tentatively identified as We used immunoblotting techniques to detect the antibody fibronectin on the basis of its size and its ability to bind in the sera of immunized rabbits, to establish the specificity collagen; and (iii) a M, 42,000 protein, tentatively identified of the antibody, and to determine the distribution of the as link protein on the basis ofits size and affinity for collagen protein in various tissues (36). Extracts (4.0 M guanidine- (38). Similar results were observed when type II collagen was HCl/0.05 M Tris HCl, pH 7.2) ofthe Swarm chondrosarcoma denatured before coupling to the Sepharose, but the protein were used as a source of antigen. After dialysis, the proteins does not bind to Sepharose-conjugated bovine serum albumin in the extract were separated by electrophoresis in a (data not shown). These studies indicate that there are three NaDodSO4/7.5% polyacrylamide gel and were transferred species of proteins in these extracts that bind to collagen, electrophoretically to nitrocellulose paper ("blots"). The including the M, 54,000 protein, which we call CBP. blots were exposed to various dilutions of antisera, and Isolation of CBP. The urea extract was chromatographed bound antibody was detected using a goat anti-rabbit IgG on a DEAE-cellulose column in 2 M urea/0.15 M NaCl/0.05 coupled to peroxidase with 4-chloro-1-naphthol as substrate M Tris HCl, pH 7.2. CBP did not bind to the column and was (36). The distribution of CBP was determined in a similar rechromatographed on a column of hydroxyapatite, using a fashion in normal serum and in 4 M guanidine HCl extracts of M as eluent. various tissues. Frozen sections of tissues were also stained gradient of phosphate buffer from 0.005-0.3 by indirect immunofluorescence according to published pro- CBP was eluted at 0.1 M phosphate (Fig. 3a). The migration the anti-CBP sera at a 1:10 dilution. position of the protein in the NaDodSO4/polyacrylamide gel cedures (37), using was identical before and after reduction (Fig. 3b), indicating RESULTS that the protein does not contain interchain disulfide bonds. CBP in Cartilage Extracts. In preliminary experiments, Tissue Distribution of CBP. The ability of CBP to bind attempts were made to identify the collagen-binding proteins collagen fibrils was utilized to isolate the protein for antibody in extracts of a cartilage-producing tumor. In these studies, the residue of cartilage matrix obtained by homogenizing tissue in 20% NaCl to remove serum and the cellular components was reextracted with a mild chaotropic solvent 200- (2 M urea). After centrifugation and dialysis against PBS, the supernatant fraction was incubated with fibrils of type II 116 _ collagen at 370C for 15 min. Collagen fibrils plus bound 93 proteins were then collected by centrifugation, washed in FIG.- 2. Type 11 couagen-sepnarose M urea PBS, dissolved in NaDodSO4/PAGE sample buffer, and - affinity chromatography. The 2 68 extract (10 ml) was dialyzed against electrophoresed (Fig. 1). In addition to the components 54- present in the original collagen preparations, fibrils incubated I1 * >PBS and chromatographed on a native type II collagen-Sepharose column. with the cartilage extract bound a protein of Mr 54,000 (Fig. 42 -- * Bound material was eluted with 6 M 1, lane 5). urea and examined by NaDodSO4/ This collagen-binding protein was also identified by affinity PAGE in the presence of a reducing chromatography using immobilized type II collagen. In these _ agent. Lanes: 1, molecular weight stan- studies, the urea extract of the tissue was dialyzed against dards; 2, 2 M urea extract; 3, bound PBS and passed over a column of native type II collagen 1 2 3 4 fraction; 4, unbound fraction. Downloaded by guest on September 29, 2021 5128 Cell Biology: Chandrasekhar et al. Proc. Natl. Acad. Sci. USA 83 (1986) a b Table 1. Tissue distribution of CBP Tissue CBP Cartilage (sternal) + + + Chondrosarcoma +++ Bone Skin Tendon + Liver Kidney 0 Nc Brain Serum Various rat tissues were extracted with 4 M guanidine hydrochlo- ride for 12 hr. The extracts were dialyzed, lyophilized, and subjected 54-'W" to electrophoresis followed by transfer to nitrocellulose. The protein CL was identified by using anti-CBP at a 1:10 dilution. hr) in comparison to type II collagen (tl2 = 2.83 hr) and in combination with proteoglycan further increased the rate (tin = 4.54 hr). The proteoglycan alone had very little effect. Fraction However, major differences were observed when the mor- phology and diameters of the fibrils were compared (Figs. 7 FIG. 3. (a) Isolation of CBP. The 2 M urea extract (100 ml) was and 8). More than 85% of type II fibrils (control) were of chromatographed on a DEAE-cellulose column, and the unbound wider diameter (>100 nm). In contrast, 99% of type II fibrils fraction was further chromatographed on a hydroxyapatite column formed in the presence of CBP plus proteoglycan were (see Materials and Methods). Bound material was eluted by a thinner in diameter (<40 nm). The fibrils formed in the gradient of phosphate buffer (0.005-0.3 M) at pH 6.8. (Inset) presence of either proteoglycan or CBP were somewhat Fractions were pooled and examined by NaDodSO4/PAGE in the larger. These results suggest that CBP binds to type II presence of a reducing agent. Proteins were visualized by Coomassie blue. Lanes: 1, molecular weight standards; 2, 2 M urea extract; 3, collagen and in combination with chondroitin sulfate peak I; 4, peak II; 5, peak III; 6, peak IV. (b) CBP before (lane 2) and proteoglycan limits the growth of collagen fibrils. after (lane 1) reduction. CBP was isolated as described in a and was examined by NaDodSO4/PAGE. DISCUSSION In recent years, a number of collagen binding proteins have preparation. Collagen fibrils were equilibrated with the ex- been identified, including fibronectin, chondronectin, tract, isolated by centrifugation, and electrophoresed, and laminin, and link protein. Fibronectin, chondronectin, and rabbits were given multiple injections with segments of gels laminin bind to one or more collagen types and also to cell containing the CBP band over a period of several weeks until surfaces and thus link cells to matrix (39). These collagen- antibody was detected by immunoblotting. binding proteins may also regulate the organization of the The antibody was found to react with a single peptide in the extracellular matrices. For example, laminin causes type IV extract, with mobility identical to the antigen (Fig. 4). A collagen to precipitate (40), and fibronectin binds to type I similar approach was used to survey for the presence of the collagen and alters its rate of fibrillogenesis (31). Antibodies protein in serum and in extracts of normal cartilage, liver, to the collagen-binding domain on fibronectin disrupt the kidney, brain, and muscle. These studies showed a positive normal arrangement in which collagen fibrils are deposited in reaction with extracts of chondrosarcoma and cartilage but cell culture (32). In addition, the link protein, along with not with other tissues (Table 1). Thus, the antibody is specific various proteoglycans, has been found to bind to fibrils of for CBP, and CBP appears to be concentrated in cartilage. type I collagen, to change the kinetics offibril formation, and Finally, immunofluorescence localization demonstrated that to change the size of the fibril formed (33, 38). Such effects CBP is present in the matrix of cartilage (Fig. 5) and not are not unexpected, given that these proteins bind tightly to present in other tissues (data not shown). collagen. Regulation ofCollagen Fibril Formation by CBP. Since CBP The size of collagen fibrils in vivo is rather constant, binds to collagen, we tested it, alone and with proteoglycan, particularly in fibrous tissues (21, 22, 24). However, the size for its effects on the fibrillogenesis of type fl collagen. CBP of fibrils formed in vitro is nonuniform. It is likely that alone (Fig. 6) increased the rate offibril formation (t12 = 3.34 additional factors normally regulate their formation. For example, using procollagen, it has been noted that fibrils of different size are generated, depending upon the order in which the procollagen precursor-specific peptides are cleaved (41). The hyaline cartilage matrix is highly organized. Its collagen fibrils are then arranged in a nonparallel fashion in which proteoglycan aggregates occupy spaces between fibrils. The factors regulating this arrangement of macromo- lecular structures are not known. Since type II collagen can FIG. 4. Specificity of antiserum to CBP. An- form large fibrils in vitro, careful regulation, presumably by tiserum (1:10 dilution) was used in a "transblot" other macromolecular constituents of the cartilage matrix, procedure against an unfractionated extract of must be involved. chondrosarcoma. The protein was identified by using peroxidase-conjugated second antibody and Cartilage contains a variety of proteins of unknown func- 4-chloro-1-naphthol substrate. Lanes: 1, molecu- tion (42). We have sought to identify putative regulators of lar weight markers; 2, chondrosarco'ma extract collagen fibril assembly by isolating collagen-binding pro- stained with amido black; 3, chondrosarcoma teins from the matrix of the Swarm chondrosarcoma, which 1 2 3 extract stained with anti-CBP. is an abundant source of newly deposited cartilage proteins Downloaded by guest on September 29, 2021 Cell Biology: Chandrasekhar et al. Proc. Nat. Acad. Sci. USA 83 (1986) 5129

FIG. 5. Indirect immunofluorescence of cartilage and the chondrosarcoma after reaction with anti-CBP. (a) Phase-contrast micrograph of unstained rib cartilage. (b) Rib cartilage stained with anti-CBP. (c) Chondrosarcoma stained with preimmune serum. (d) Chondrosarcoma stained with anti-CBP. (35, 43). Although the cells in this chondrosarcoma are transformed, they produce apparently normal cartilage ma- trix macromolecules and deposit them in a regular fashion. Matrix components were solubilized in a chaotropic solvent, and their affinity for type II collagen was tested in physio- logical solvents. CBP was the major component found to bind to type II collagen , although proteins tentatively identified as fibronectin and link protein also showed some binding. Immunization of rabbits with CBP produced an antibody that recognized CBP in extracts of the chondrosar- coma and of normal cartilage but not in extracts of various

11 + CBP

C,, II~~~~~~1+PG 0 0.5- _O I + PG +CBP 0D

Time, hr FIG. 6. Effect of CBP and proteoglycan on type II collagen fibril formation. Type II collagen (0.6 mg) was mixed with CBP (0.05 mg) and/or chondroitin sulfate proteoglycan (0.1 mg) and incubated at 0C for 10 min in a final volume of 1.0 ml. The solutions were quickly brought to 350C and monitored for changes in optical density at 314 nm, and the turbidity curves were recorded. The maximal optical FIG. 7. Visualization of the effect of CBP and proteoglycan on density recorded was for type II collagen alone (OD = 2.2). The other type II collagen fibril formation. Samples were withdrawn aftera 6-hr OD values are expressed relative to those obtained for type II incubation at 35°C and examined in the electron microscope. (a) collagen alone, which is considered as 1. The tln for fibril formation Type II collagen alone. Most of the fibrils are large and cross- for type II collagen alone is 3.34 hr; for type II collagen plus striated. (b) Type II collagen plus CBP and proteoglycan. All of the proteoglycan (PG), 3.63 hr; for type II collagen plus CBP, 2.83 hr; fibrils are narrow throughout the field; the fibrils show a slight and for type II collagen plus CBP and PG, 4.43 hr. cross-striation. Some branching is seen. (Bar = 100 am.) Downloaded by guest on September 29, 2021 5130 Cell Biology: Chandrasekhar et al. Proc. NatL Acad. Sci. USA 83 (1986) 5. Gregory, J. D. (1973) Biochem. J. 133, 353-386. 6. Heinegard, D. K. & Hascall, V. C. (1974) J. Biol. Chem. 249, 4250-4256. 7. Baker, J. & Caterson, B. (1977) Biochem. Biophys. Res. Commun. 77, 1-10. 8. Hascall, V. C. (1981) in Biology of Carbohydrates, ed. 0 Ginsburg, V. (Wiley, New York), Vol. 1, pp. 1-49. 50 9. Poole, A. R., Reiner, A., Tang, L.-H. & Rosenberg, L. C. (1980) J. Biol. Chem. 255, 9295-9305. C. 10. Poole, A. R., Pidoux, I., Reiner, A. & Rosenberg, L. (1982) J. U- Cell Biol. 93, 921-937. 11. Miller, E. J. (1971) Biochemistry 10, 1652-1659. 12. Mayne, R. & von der Mark, K. (1982) in Cartilage, ed. Hall, B. K. (Academic, New York), Vol. 1, pp. 181-214. 13. Workman, L., Chandrasekhar, S. & Balian, G. (1984) J. Cell. 0 Biochem., Suppl. 8B, 286 (abstr.). Type II Type 11 Type 11 Type 11 14. Bornstein, P. & Sage, H. (1980) Annu. Rev. Biochem. 49, + CBP + PG + PG + CBP 957-1003. 15. Gross, J. & Kirk, D. (1958) J. Biol. Chem. 223, 355-360. FIG. 8. Diameter oftype II collagen fibrils formed in the presence 16. Wood, G. C. (1960) Biochem. J. 75, 605-612. or absence of CBP and proteoglycan (PG). Fibril diameters were 17. Keech, M. K. (1961) J. Biophys. Biochem. 9, 193-209. measured after a 6-hr incubation at 350C. Open bars represent fibrils 18. B. R. A. & K. A. J. <40 nm in solid bars fibrils 100-250 nm Williams, R., Gelman, Piez, (1978) Biol. diameter; represent in Chem. 253, 6578. diameter. Number of fibrils scored was as follows: for type II 19. Trelstad, R. L. & Silver, F. H. (1981) in Cell Biology of collagen alone, 84; for type H collagen plus for type II CBP, 114; Extracellular Matrix, ed. Hay, E. D. (Plenum, New York), pp. collagen plus PG, 108; for type II collagen plus PG and CBP, 105. 179-216. 20. Kuhn, K. (1982) Collagen Relat. Res. 2, 61-80. other tissues or in serum. Taken together, these results 21. Trelstad, R. L. (1982) Cell 28, 197-198. suggest that CBP is a cartilage-specific protein. 22. Gross, J. & Bruns, R. R. (1984) in The Role ofExtracellular A possible function for this protein in the regulation of Matrix in Development, ed. Trelstad, R. L. (Liss, New York), collagen fibrillogenesis was sought based on its interaction pp. 479-512. with type II collagen. In confirmation ofpast studies (26), we 23. Birk, D. E. & Silver, F. H. (1984) Arch. Biochem. Biophys. found that purified type II collagen reconstitutes into large, 235, 178-185. 24. Piez, K. A. (1984) in Extracellular Matrix Biochemistry, eds. well-ordered fibrils in vitro. Addition of either CBP or Piez, K. A. & Reddi, A. H. (Elsevier, New York), pp. 1-40. chondroitin sulfate proteoglycan gave rise to a mixture of 25. Stark, M., Miller, E. J. & Kuhn, K. (1972) Eur. J. Biochem. large plus small fibrils. However, addition of CBP and 27, 192-196. proteoglycan together with type II collagen resulted in fibrils 26. Lee, S. L. & Piez, K. A. (1983) Collagen Relat. Res. 3, 89-103. of uniformly small diameter (24 ± 8 nm). These results 27. Toole, B. P. & Lowther, D. A. (1968) Biochem. J. 109, 857-866. suggest that CBP in combination with the cartilage-specific 28. Obrink, B., Laurent, T. C. & Carlsson, B. (1972) FEBS Lett. collagen and proteoglycan generates small fibrils character- 56, 166-169. istic of hyaline cartilage. Thus, it is likely that multiple 29. Oegema, T. R., Laidlaw, J., Hascall, V. C. & Dziewiatkowski, interactions are involved in the formation of the cartilage D. C. (1975) Arch. Biochem. Biophys. 170, 698-709. 30. Snowden, J. M. & Swann, D. A. (1980) Biopolymers 19, matrix. Various data suggest that the formation of other 767-780. extracellular matrices also requires several interacting com- 31. Kleinman, H. K., Wilkes, C. M. & Martin, G. R. (1981) Bio- ponents. For example, ternary complexes of this type have chemistry 20, 2325-2330. been noted before with type I collagen, fibronectin, and 32. McDonald, J. A., Kelley, D. G. & Brockelmann, J. J. (1982) J. heparan sulfate (44). In addition, type IV collagen, laminin, Cell Biol. 92, 485-493. and heparan sulfate proteoglycan likely interact (40) to form 33. Chandrasekhar, S., Kleinman, H. K., Hassell, J. R., Martin, networks of 3- to 8-nm-wide cords which are concentrated in G. R., Termine, J. D. & Trelstad, R. L. (1984) Collagen Relat. the lamina densa part of basement membrane a Res. 4, 323-338. (45). Thus, 34. Palmoski, M. J. & Goetinck, P. F. (1972) Proc. Natl. Acad. general feature determining matrix structures may be that Sci. USA 69, 3385-3389. collagens interact with other matrix components. In the case 35. Smith, B. D., Martin, G. R., Miller, E. J., Dorfman, A. & ofcartilage, mutants lacking chondroitin sulfate proteoglycan Swarm, R. (1975) Arch. Biochem. Biophys. 166, 181-186. have a matrix containing thin collagen fibrils, suggesting that 36. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl. proteoglycan is not essential (34). On the other CBP Acad. Sci. USA 76, 4350-4354. hand, 37. Foidart, J. M., Berman, J. J., Paglia, L., Rennard, S. I., Abe, S., may well bind to proteoglycan as well as to collagen and could Perantoni, A. & Martin, G. R. (1980) Lab. Invest. 42, 525-532. link these apparently independent aggregates into one com- 38. Chandrasekhar, S., Kleinman, H. K. & Hassell, J. R. (1983) J. mon structure. Biol. Chem. 258, 6226-6231. 39. Kleinman, H. K., Klebe, R. & Martin, G. R. (1981) J. Cell We thank Gertrude Goping and William Johnson for technical Biol. 88, 473-485. assistance and Irma Burke for preparing the manuscript. This work 40. Kleinman, H. K., McGarvey, M. L., Hassell, J. R. & Martin, was supported by grants from the intramural division of the National G. R. (1983) Biochemistry 22, 4969-4974. Institute of Dental Research and the Kroc Foundation. 41. Miyahara, M., Hayashi, K., Burger, J., Tanzawa, K., Njieha, F. K., Trelstad, R. L. & Prockop, D. J. (1984) J. Biol. Chem. 1. Muir, H., Bullough, D. & Maroudas, A. (1970) J. Bone J. Surg. 259, 9891-9898. Br. Vol. 52, 554-563. 42. Paulsson, M. & Heinegard, D. (1984) Collagen Relat. Res. 4, 2. Hukins, D. W., Knight, D. P. & Woodhead-Galloway, J. 219-230. 43. Angerman, K. & Barrach, H.-J. (1979) Anal. Biochem. 94, (1976) Science 194, 622-624. 253-258. 3. Paulsson, M. & Heinegard, D. K. (1979) Biochem. J. 183, 44. Hayman, E., Oldberg, A., Martin, G. R. & Ruoslahti, E. 539-545. (1982) J. Cell Biol. 94, 28-35. 4. Hascall, V. C. & Sajdera, S. W. (1969) J. Biol. Chem. 244, 45. Laurie, G. W., Leblond, C. P., Inoue, S., Martin, G. R. & 2389-2396. Chung, A. (1984) Am. J. Anat. 169, 463-481. Downloaded by guest on September 29, 2021