THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 2, Issue of January 15, pp. 1030-1036,1988 Printed in U. S. A.

Biosynthesis and Cell-free Translation of Swarm Rat Chondrosarcoma and Bovine Cartilage Link Proteins*

(Received for publication, August 4, 1987)

Thomas M. Hering4 and Linda J. Sandells From the Department of Biochemistry, Department of Orthopaedic Surgery, Rush-Presbyterian-&. Luke’s Medical Center, Chicago, Illinois 60612

In cartilage, link protein(s) (LP) stabilize proteogly- can aggregates via their specific association with hy- aluronic acid and proteoglycan monomers. Two major link glycoproteins are produced in bovine articular cartilage, designated LP1 (49.5 kDa) and LP2 (44.0 kDa), whereas rat chondrosarcoma produces a single link protein species similar in size to bovine LP2. Although multiple link proteins differ to a significant degree in carbohydrate content, it is not known whether they arise from variable glycosylation of a single common protein core or from complete glycosyl- ation of different protein cores. Biosynthesis of these molecules has been studied under conditions where differences generated by N-linked glycosylation would not be evident. Link proteins were immunoprecipitated 1) from cell-free translation products of total cellular and size fractionated RNA and 2) from cell lysates and medium of cultured chondrocytes using short term ra- dioactive labeling of the protein in the presence and absence of tunicamycin. A 42-kDa link protein precur- sor is synthesized by cell-free translation of either rat chondrosarcoma or bovine chondrocyte mRNA. An ap- parently single 41.5-kDa link protein is synthesized with inhibition of N-linked glycosylation by tunica- mycin, whereas LP1 and LP2 are the mature products of cultured bovine chondrocytes. The size range of translatable rat chondrosarcoma LP mRNA is 4.0-5.5 kilobase pairs and bovine LP mRNA is 3.0-4.5 kilo- base pairs, both much larger than required to encode the link protein molecule. These results suggest that a single link protein precursor gives rise to multiple fully glycosylated forms and that link protein is not synthe- sized as a significantly larger “pro” form.

The predominant components of the cartilage extracellular matrix are type I1 collagen fibers and chondroitin sulfate proteoglycan aggregates composed of proteoglycan monomers, hyaluronic acid, and link proteins. Link proteins stabilize the association between proteoglycan monomers and hyaluronic

Grants AM 34142, AR 36694 (to L. J . S.), and AM0 7375-07 (to T. *This research was supported by National Institutes of Health

M. H.), National Institutes of Health Postdoctoral Fellowship Award F23AR07772 (to T. M. H.), a March of Dimes Birth Defects Basic Research Grant (to L. J . S.), and grants from the Rush University Committee on Research (to T. M. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$Present address: Dept. of Orthopaedics RK-10, University of Washington, Seattle, WA 98195.

§Present address: Dept. of Orthopaedics RK-10, University of Washington, Seattle, WA 98195. To whom reprint requests should t e addressed.

acid (Hardingham, 1979; Tang et al., 1979; Kimura et al., 1979; 1980), influence the dimensions of the aggregate (Buckwalter et al., 1984), and consequently may play an important func- tional role in the organization and maintenance of extracel- lular matrix architecture in cartilage. Link proteins can also bind native non-cartilage collagen types I and I11 (Chandra- sekhar et al., 1983) and have been identified in aorta (Gardell et al., 1980; Vijayagopal et al., 1985) and as products of cultured synovial fibroblasts (Fife et al., 1985).

In most cartilaginous tissues, multiple link proteins are present varying in their electrophoretic mobility, carbohy- drate content (Baker and Caterson, 1979), and binding prop- erties (Choi et al., 1985). Although the different molecular forms of link protein are immunologically related and have no detectable differences in amino acid composition (Bonnet et al., 1978; Baker and Caterson, 1979), it has not been determined whether the multiple forms of link protein are due to differences in glycosylation of a common protein core or of different but closely related proteins. To date, little is known concerning the molecular basis of link protein heter- ogeneity or the control of multiple link protein synthesis. To begin to address this problem, we have examined the biosyn- thesis of link proteins under conditions where differences due to glycosylation and the generation of additional species due to proteolytic degradation would not be evident. The biosyn- thesis of link proteins was compared in cells from two tissues, one of which produces only one mature link protein, the Swarm rat chondrosarcoma, and one which produces multiple link proteins, bovine articular cartilage. The primary gene products of link protein mRNA were identified in both sys- tems. Bovine articular cartilage chondrocytes in culture were used as a model system to characterize normal link protein biosynthetic products as well as the link proteins synthesized in the presence of an inhibitor of Asn-linked carbohydrate addition.

EXPERIMENTAL PROCEDURES

Materials-RCS’ cells were provided by Dr. James Kimura, Rush- Presbyterian-St. Luke’s Medical Center. DNA probes for 18 and 28 S rRNA were provided by Dr. Ira Wool, The University of Chicago. The rabbit reticulocyte lysate translation system was purchased from Bethesda Research Laboratories. [Y3]Methionine (1300 Ci/mmol) was purchased from Amersham Corp. and [3H]leucine (140.8 Ci/ mmol) from Du Pont-New England Nuclear. Tunicamycin and pro- tein A-Sepharose were obtained from Sigma. Antiserum R13 was obtained from Dr. Bruce Caterson, University of West Virginia, and Dr. James Christner, University of Alabama at Birmingham. Mono- clonal antibodies 8A4, 8C1, and 3B1 were provided by Dr. Bruce Caterson, University of West Virginia. Nonradioactive protein mo-

‘The abbreviations used are: RCS, Swarm rat chondrosarcoma; LP, link protein; BAC, bovine articular chondrocyte; mAb, monoclo- nal antibody; NRS, normal rat serum; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; HEPES, 4-(2-hydroxy- ethyl)-1-piperazineethanesulfonic acid; kb, kilobase pair(s).

1030

Biosynthesis and Cell-free Translation of Link Protein 1031

lecular weight standards were obtained from Bio-Rad and I4C-labeled protein molecular weight standards from Bethesda Research Labo- ratories.

Cell Culture-Bovine articular chondrocytes were isolated from the metacarpophalangeal joints of adult cows and plated at 13 X lo6 cells/ 100-mm culture dish as described by Kuettner et al. (1982). Cultures were maintained for 5 days in F12 medium supplemented with 10% fetal calf serum and 25 pg/ml ascorbic acid in a humidified atmos- phere of 5% COa and 95% air at 37 "C.

RNA Isolation-RCS cells thawed from storage at -70 "C and cultured BAC were washed with Hanks' balanced salts solution. Extracellular matrix was removed by digestion with 0.4% collagenase/ 0.25% trypsin at 37 "C until single cells could be observed microscop- ically. The cells were dispersed in 0.1-0.5 ml of 8 M guanidine hydrochloride/l mM dithiothreitol/20 mM sodium acetate, pH 6.0, for each lo6 cells and homogenized in a Wheaton Dounce homoge- nizer. Cell debris was removed by centrifugation at 3000 X g for 15 min, and nucleic acids were precipitated from the supernatant with 0.5 volumes of 100% ethanol. RNA was isolated by sequential precip- itations from 8 M guanidine hydrochloride/l mM dithiothreitol/20 mM sodium acetate, pH 6.0/20 mM EDTA with 0.5 volumes of 100% ethanol and from 10 mM HEPES, pH 7.2/0.3 M sodium acetate with 2.5 volumes of 100% ethanol. The RNA extraction and purification were a modification of the method of Strohman et al., (1977) as previously described (Sandell and Daniel, 1987).

Sucrose Gradient Fractionation of RNA-Total cellular RNA iso- lated from either RCS cells or bovine chondrocytes was size-fraction- atea on a gradient of sucrose by the method of Sawhney? RNA (600 pg) was dissolved in 5 mM Tris-HC1, pH 7.4, containing 250 mM NaC1, 0.5 mM EDTA, and 0.1% SDS. Samples were denatured by heating for 5 min at 65 "C and cooled in an ice-water bath. Denatured samples were layered onto the surface of a 16-ml, 10-35% sucrose gradient in 5 mM Tris-HC1, pH 7.4, containing 250 mM NaCl, 0.5 mM EDTA, and 0.1% SDS and were centrifuged using a Beckman SW 28 rotor at 27,000 rpm for 15.5 h at 21 "C. Immediately following the centrifugation run, 260-pl fractions were collected using a Buchler Auto Densi-Flow I1 in line with a peristaltic pump. Gradient fractions were ethanol precipitated and were reprecipitated from a large volume of 10 mM HEPES, pH 7.2, to ensure adequate removal of SDS. Following centrifugation, the RNA pellets were washed sequentially with 70% ethanol and 100% ethanol and then dried 15 min at room temperature. Finally, the pellets were redissolved in 20 pl of sterile distilled water and frozen at -20 "C.

RNA Electrophoresis and Northern Blot Analysis-RNA was elec- trophoresed on agarose gels containing formaldehyde as described by Maniatis et al. (1982), with modifications. Following electrophoresis, RNA was transferred from the agarose gel to nitrocellulose as de- scribed by Thomas (1980). Filters were prehybridized, hybridized with 32P-labeled DNA probes, washed, and autoradiographed as de- scribed by Alwine et al. (1979).

Cell-free Transslation-RNA was translated in vitro as described by Pelham and Jackson (1976) using a micrococcal nuclease-treated rabbit reticulocyte lysate translation system incorporating 13'S]me- thionine at a concentration of 1.5 pCi/pl reaction volume. The typical translation reaction (40 pl) contained 13.3 pl of nuclease-treated reticulocyte lysate, 82 mM KOAc, 1.16 mM MgC12, 50 mM of all amino acids except methionine, 60 pCi of [35S]methionine, and 100-200 pg/ ml RNA.

cin in 1.5 p1 of Me2S0 were added to each of three, 5-day BAC Tunicamycin Treatment of BAC Cultures-Dilutions of tunicamy-

cultures such that the final tunicamycin concentrations were 0.3,3.0, and 30.0 pg/ml of medium. Control cultures were left untreated or were treated with 1.5 pl of MezSO only. Plates were incubated for 6 h. Cells were then labeled for 17 h with [3H]leucine at 50 pCi/ml of medium prior to immunoprecipitation of link protein from cell lysates and medium.

modification of the method of Kimura et al. (1987). To 40 p l of cell- Immunoprecipitation-Samples were immunoprecipitated using a

free translation mixture was added 10 pl of 250 mM HEPES, pH 8.0, containing 10% lithium dodecyl sulfate, 50 mM disodium EDTA, 5 mM phenylmethylsulfonyl fluoride, and 1.8 mM pepstatin A. Samples were reduced by heating at 100 "C for 5 min following addition of 1.3 p1 of 0.4 M dithiothreitol (to 10 mM), and cooled samples were alkylated by incubation in the dark at room temperature for 2 h following addition of 1.3 p1 of 0.8 M iodoacetamide (to 20 mM).

Samples were diluted by addition of 1.0 ml of 50 mM HEPES, pH 7.4, containing 1% w/v Nonidet P-40,1% (w/v) sodium deoxycholate, 0.15 M NaC1, 10 mM EDTA, 0.1 M 6-aminohexanoic acid, 10 mM N - ethylmaleimide, 5 mM benzamidine-HC1, and 1 mM phenylmethyl- sulfonyl fluoride. Samples were centrifuged, and 1.0 ml of the super- natant was transferred to another tube. Polyclonal antiserum or diluted monoclonal antibody (10-12.5 pl) was added, and samples were left at 4 "C overnight. To each immunoreaction mixture was then added 216 p1 of a 3% suspension of protein A-Sepharose in washing buffer (the same composition as the dilution buffer above but with the addition of SDS to a final concentration of 0.1%). Samples were then placed on a rotating mixer for 3 h at 4 "C. Tubes were then centrifuged at 800 X g for 5 min at 4 "C. The initial supernatant was drawn off, and the protein A-Sepharose pellets were washed six times each with 1 ml of washing buffer and four times with 10 mM Tris-HC1, pH 6.8. After the final wash, the pellets were frozen at -20 'C until SDS-PAGE analysis.

Antiserum and Monoclonal Antibodies-The monoclonal antibody (mAb) 8A4 was used to immunoprecipitate link protein from cell-free translation products of rat chondrosarcoma. Although rat chondro- sarcoma LP2 was used as the antigen for the production of mAb 8A4, the immunoglobulin recognizes LP1,2, or 3 isolated from rat, bovine, and chick hyaline cartilages. The antibody determinant is believed to be a part of a conserved sequence of the link protein polypeptide common to all link proteins. The epitope recognized by mAb 8A4 is believed to reside in the carboxyl-terminal portion of the link protein molecule (Caterson et al., 1985). As a negative control, identical samples were immunoprecipitated with a different monoclonal anti- body designated 8C1. mAb 8C1 recognizes an epitope in the native hyaluronic acid binding region of RCS chondroitin sulfate proteogly- can core protein (330 kDa). Although nascent chondrosarcoma pro- teoglycan core protein is expected to be present in the cell-free translation products in reduced form, it would not be recognized by mAb 8C1? Therefore, any bands observed in the mAb 8C1 immuno- precipitate may be considered nonspecific. A polyclonal antiserum designated R13 was used to immunoprecipitate link protein from cell- free translation products of bovine mRNA or from cultured bovine chondrocyte lysates and medium. Antiserum R13 is a rabbit anti- serum raised against link proteins from bovine nasal cartilage which reacts predominantly with determinants in the amino-terminal half of the molecule (Baker et al., 1983). Identical samples were immu- noprecipitated with normal rabbit serum as a negative control. The same link protein bands were seen when R13 and mAb 8A4 were used in parallel immunoprecipitations. SDS-PAGE/Fluorography-SDS-polyacrylamide gel electropho-

resis was performed according to Laemmli (1970). Samples of the cell-free translation products (1 p1/40 pl) were added to aliquots of sample buffer containing dithiothreitol. The washed protein-A-Seph- arose pellets were lyophilized and suspended in aliquots of sample buffer with or without dithiothreitol. In some samples it was advan- tageous to allow the IgG to migrate in its unreduced form. Samples were heated at 100 "C for 5 min, cooled, and centrifuged through Bio- Rad econo-columns (catalog number 731-1550) in order to remove the Sepharose beads and to ensure uniformity of sample recovery. Duplicate or triplicate immunoprecipitated samples were pooled at this step whenever pooling was necessary to visualize link protein. Molecular weights were calculated by comparison to migration of "C- labeled or unlabeled standard proteins run on each gel.

Samples were electrophoresed on 10.5% separating gels 8.0 cm long X 0.15 cm in thickness with a 4.0% stacking gel, 1.0 cm long. Samples were run for 2.5 h at 35 mA/gel. Following electrophoresis, the lanes containing nonradioactive molecular weight standards were cut from the gel, electroblotted onto nitrocellulose, and stained with Amido Black. The remaining portion of the gel was fixed overnight in isopropanol/acetic acid/water (2.5/1/6.5) and prepared for fluorog- raphy (Bonner and Laskey, 1974).

RESULTS

Translation of Rat Chondrosarcoma mRNA and Immuno- precipitation of Nascent Link Protein-The link protein pre- cursor immunoprecipitated from cell-free translation products of RCS mRNA is an apparently single 42-kDa protein. A size difference is observed between the link protein precursor from cell-free translation and the fully processed link protein pro-

R. s. Sawhney, personal communication. B. Caterson, personal communication.

1032 Biosynthesis and Cell-free Translation of Link Protein

I -

30 ' 1 1 I

0 20 40 60 80 ~-

Migration Distance (mm)

FIG. 1. Immunoprecipitation of link protein precursor from cell-free translation products of RCS mRNA; size comparison with LP synthesized by RCS chondrocytes. Cell-free translation products directed hy RCS mRNA were labeled by incorporation of [%]methionine, reduced and alkylated, and immunoprecipitated with mAh 8A4 (1/4000 dilution). Following electrophoresis, the gel was sliced into 1-mm sections, and radioactivity was determined by liquid scintillation counting. The link precursor migrated as a single band with an apparent molecular mass of 42 kDa, as compared to the migration of standard proteins. Link protein immunoprecipitated from [RH]leucine-laheled RCS culture medium (arrow) migrates with an apparent molecular mass of 44.5 kDa.

duced by RCS cells in culture (Fig. 1). The cellular LP denoted by the arrow in Fig. 1 migrates more slowly (44.5 kDa) than the cell-free translation product (42 kDa). Slower migration of the RCS cell-derived link protein is expected due to the presence of a single Asn-linked carbohydrate chain (Neame et al., 1986). In another experiment, RCS translation products were immunoprecipitated with mAb 8A4, specific for link protein, and with mAb 8C1, as a negative control. The RCS translation products, mAb 8A4 immunoprecipitates using 1/ 400 and 1/2000 dilutions of antibody, and the mAb 8C1 immunoprecipitates using the same dilutions of antibody were subjected to SDS-PAGE and fluorography (Fig. 2). Lane T shows the full range of translation products of rat chondro- sarcoma mRNA. The mAb 8A4 immunoprecipitates (lanes 2 and 2) show a single prominent band with an apparent mo- lecular mass of 42.0 kDa which is not present in the mAb 8C1 immunoprecipitates (lanes 3 and 4) . The intensity of the 42- kDa band decreases when a higher dilution of antibody is used.

Translation of Bovine Articular Chondrocyte mRNA and Immunoprecipitation of Nascent Link Protein-BAC mRNA was used to direct protein synthesis in the rabbit reticulocyte lysate translation system. The link protein cell-free transla- tion product was immunoprecipitated with the polyclonal antiserum R13 and with normal rabbit serum (NRS). The R13 immunoprecipitate from bovine chondrocyte mRNA cell- free translation products was subjected to SDS-PAGE and fluorography (Fig. 2). The R13 immunoprecipitate (lane B) shows a single prominent band with an apparent molecular mass of 42.0 kDa which is not present in the NRS immuno- precipitate (not shown). The band at 70 kDa is not seen when the link protein precursor is immunoprecipitated with the mAb 8A4 and is considered to be an artifact. Unreduced IgG (translation products were reduced and alkylated prior to immunoprecipitation) migrates near the top of the lane, dis- placing the labeled background translation products. In this figure the R13 immunoprecipitate from the BAC mRNA translation products was electrophoresed in a lane adjacent to the mAb 8A4 immunoprecipitate from the RCS mRNA translation products. Although both are estimated to be 42- kDa proteins by comparison to molecular weight standards, the bovine precursor migrates slightly more rapidly.

TO confirm the identity of the bovine link protein precursor, three bovine chondrocyte RNA translation mixtures were

B 1 2 T 3 4

200.0-

9 7.4- 6 8.0-

- .,' e ' -

43.0- P

25.7-

18.4-

-a FIG. 2. Immunoprecipitation of link protein precursor from

cell-free translation products of RCS rnRNA; size comparison with precursor from BAC mRNA. [%]Methionine-labeled im- munoprecipitates were analyzed by SDS-PAGE on 10.5% gels and hands were visualized by fluorography. Lane T shows the translation products of RCS mRNA. Lanes I and 2 show mAb 8A4 immunopre- cipitates of reduced and alkylated RCS translation products using I/ 400 and 1/2000 dilutions of antihody. A higher dilution of antibody results in diminished intensity of the 42-kDa band. Lanes 3 and 4 contain mAb 8C1 (nonspecific) immunoprecipitates of identical RCS translates using 1/400 and 1/2000 antibody dilutions, showing no 42- kDa hands. Lane R shows an antiserum R13 immunoprecipitate of reduced and alkylated BAC translation products run on the same gel to demonstrate a slight size difference between the RCS and the BAC link precursor. To prevent distortion of the link protein bands, samples were run without dithiothreitol, permitting IgC to migrate in its unreduced form at the top of each lane.

immunoprecipitated with 1/1000, 1/200, and 1/100 dilutions of R13 antiserum, respectively (Fig. 3, lanes 2-3). As a control, three identical translation mixtures were immunoprecipitated with the same dilutions of NRS (Fig. 3). Immunoprecipitates were subjected to SDS-PAGE and fluorography. In this ex- periment, samples were reduced prior to electrophoresis. As shown previously, the R13 immunoprecipitate showed a prom- inent band with an apparent molecular mass of 42.0 kDa, the intensity of which increased with increasing amounts of the specific antiserum (lanes 2-3). The intensity of the back- ground bands remained essentially the same in each lane. Since the IgG eluted from the protein A-Sepharose was re- duced prior to sample application, it migrates as 50-kDa heavy chains and 23.5-kDa light chains. The increasing quantity of the IgG heavy chain in each lane from 1-3 can be seen to increasingly distort the '?+labeled background proteins mi- grating in the 50-kDa region. There was no 42-kDa band visible in the immunoprecipitates using increasing amounts of NRS. However, increasing distortion of labeled background proteins by the unlabeled IgG heavy chain eluted from the protein A-Sepharose was observed. Although the single nar- row bands observed in the RCS and BAC immunoprecipitates imply single LP precursor proteins, the possibility of co- migrating multiple proteins cannot yet be eliminated.

Immunoprecipitation of Link Protein from Tunicamycin- treated BAC Cultures-Link proteins immunoprecipitated from cell lysates and medium of untreated bovine articular

Biosynthesis and Cell-free Translation of Link Protein 1033

I

FIG. 3. Immunoprecipi ta t ion of the l ink protein precursor from cell-free translation products of BAC mRNA using in- creasing amounts of antiserum. Imzo 7‘ shows the translation products of HAC mRNA. Imws 1-3 contain antiserum R13 immu- noprecipitates of BAC translation products using 1/1000, 1/200, and 1/100 dilutions of antiserum, respectively, showing increasing inten- sity of the 42-kDa band (arrow) with increasing antiserum. Lanes 4- 6 show nonimmune rabbit serum control immunoprecipitates of BAC translation products using 1/1000, 1/200, and 1/100 dilutions of serum. Nonspecific bands immediately above the link precursor are the same in immune and nonimmune serum immunoprecipitates. lmmunoprecipitation of the higher molecular weight band visible in lanes 1-3 was variable between lots of nonimmune serum and is considered to be nonspecific. Samples were reduced prior to electro- phoresis causing the IgC heavy chain to distort nonspecific bands.

chondrocyte cultures were similar to those reported for bovine articular cartilage (Treadwell et al., 1980a). Two major bands were visible in R13 immunoprecipitates of untreated chondro- cyte lysate (Fig. 4, lane C l ) and medium (lane M I ) migrating with apparent molecular masses of 49.5 and 44.0 kDa. The lower relative intensity of the bands in lanes C l and MI is due to the loading of only half the sample on this gel. Chon- drocytes were also grown in the presence of tupicamycin in order to determine whether a link protein precursor similar in size to the cell-free translation product could be obtained when N-linked glycosylation was inhibited. Tunicamycin in- hibits the formation of the dolichol intermediate in the bio- synthesis of N-asparaginyl-linked oligosaccharides (Struck and Lennarz, 1980; Takatsuki and Tamura, 1971). Link pro- tein was immunoprecipitated from bovine articular chondro- cyte cultures that were treated with increasing doses of tuni- camycin prior to labeling with [‘lH]leucine. Lysates and me- dium from Me2S0 (tunicamycin solvent) -treated cultures (lanes C2 and M2) show the same major species. A 41.5-kDa LP3 band is visible in the immunoprecipitate from the me- dium of Me,SO-treated cultures (lane M 2 ) and is faintly visible in the cell lysate (lane CZ). Lysate and medium im- munoprecipitates of cultures treated with 0.3 pg/ml tunica- mycin (lanes C3 and M 3 ) show an increase in the relative intensity of a band migrating with an apparent molecular mass of 41.5 kDa, similar in size to the LP3 band. Lysate and medium immunoprecipitates of cultures treated with 3.0 and 30.0 pglml tunicamycin (lanes C4 and C5, M4 and M5) show complete elimination of the 49.5- and 44.0-kDa bands and a further increase in the intensity of the 41.5-kDa band. The

C M 1 2 3 4 5 1 2 3 4 5

4 9.5 z

44.0; 41.5

FIG. 4. Immunoprecipi ta t ion of l ink protein from untreated and tunicamycin-treated cultured BAC cell lysate and me- dium. Antiserum R13 was used to immunoprecipitate link protein from untreated cell lysate and medium (lanes C1 and M I ) , lysate and medium treated with Me2S0 (tunicamycin solvent) only (lanes CZ and MZ), treated with 0.3 pg/ml tunicamycin (lanes C3 and M3), 3.0 pg/ml tunicamycin (lanes C4 and M 4 ) , and 30.0 pg/ml tunicamycin (lanes C5 and M5). Inhibition of N-linked glycosylation by tunica- mycin results in a dose-dependent loss of the 49.5- and 44.0-kDa bands (LP1 and LP2) and corresponding increase in a 41.5-kDa band slightly smaller in size than the link protein precursor from cell-free translation (42 kDa).

light bands visible in lanes Cl-C5 above the link protein bands are apparently due to incomplete washing of other cell lysate proteins from the protein A-Sepharose beads. Inhibi- tion of N-glycosylation results in the prevention of synthesis of LP1 and LP2. Following tunicamycin treatment, the cel- lular as well as the secreted product show a marked increase in electrophoretic mobility similar to that of LP3. Comparison of the sharply defined tunicamycin product band with the more diffuse LP1 and LP2 bands seen prior to treatment suggests a loss of microheterogeneity in the link protein following treatment to inhibit N-glycosylation. It is apparent that inhibition of N-glycosylation does not prevent secretion, since abundant non-N-glycosylated link protein can be im- munoprecipitated from the medium as well as the cell lysate.

The LP3-sized link protein produced following tunicamycin treatment is probably not identical to the LP3 seen in the untreated overnight labeled cultures, which may have arisen from partial extracellular degradation. We have recently ob- served that although LP3 can be immunoprecipitated from the medium of cultured chondrocytes following long-term labeling, it cannot be detected in the cell lysates following short-term labeling.4

Analysis of Sucrose Gradient Fractionated RNA and Im- munoprecipitation of the Link Protein Precursor from Cell- free Translation Products of Fractionated RCS and BAC mRNA-Size-fractionated RCS and BAC total cellular RNA was prepared by sucrose gradient centrifugation for determin- ing the size of translatable link protein mRNA. RCS (Fig. 50) and BAC total cellular RNA were fractionated on 10-35% sucrose gradients. The distribution of BAC total cellular RNA was identical to that of the RCS. Two major peaks of absorb- ance a t 260 nm were evident, representing 18 S (1900 bases) and 28 S (4700 bases) rRNA. Because the RNA had been

‘ T. M. Hering and L. J. Sandell, manuscript in preparation.

1034 Biosynthesis and Cell-free Translation of Link Protein

RNA FRACTIONATION subjected to only mild denaturing conditions, the quality of

ul( I I ) - - 3 5 the size fractionation was determined by examining the dis- tribution of markers across the gradient. The distribution of

- ribosomal RNAs and two mRNAs of known size was deter-

0 blot analysis of sucrose gradient fractions, hybridizing to DNA

P 1.0 - 7 (indicated by the bar labeled 18 S Agg.). This may be due to

- 15 encoding the small cartilage proteoglycan core protein (43

2.0

s - 30 mined. The distribution of rRNA was verified by Northern

0 25 2 probes for 18 and 28 S rRNA. The 18 S probe hybridized with g

f a small quantity of RNA co-fractionating with rRNA 228 S

high molecular weight aggregates of 18 S rRNA or high molecular weight 18 S rRNA precursors. Functional mRNA

kDa) peaked at the theoretical expected size (1200-2900) bases) in the BAC RNA gradient.s Consequently, functional mRNA does not appear to be aggregated. Rat chondrosarcoma gradient blots were hybridized with a probe for mRNA coding

a - 2 0 -

0 10 20 30 40 50 FRACTION NO.

RAT ! ~ o p n ~ o ~ ! o ~ o ~ BOVINE POOLS 1- 94d4 for the al(I1) chain of type I1 collagen. Hybridization was

A B C T 1 2 3 4 5 6 T 1 2 3 4 5 6

A B C T 1 2 3 4 5 6 T 1 2 3 4 5 6

FIG. 5. Immunoprecipitation of link protein precursor f rom cell-free translation products of sucrose gradient-fraction- a t ed RNA. RCS and BAC total cellular RNA were size-fractionated on a 1045% sucrose gradient, and the pattern for RCS is shown (a) . The fractionation pattern of bovine chondrocyte total RNA was identical. Peaks represent 18 S (1900 bases) and 28 S (4700 bases) rRNA, confirmed by Northern blot analysis of fractions using a specific probe, which also revealed that either high molecular weight aggregates or large precursors of 18 S rRNA (indicated by the bar labeled 18 S Agq.) fractionated in the region 228 S. As a high molecular weight RNA marker, rat RNA fractions containing type I1 collagen mRNA (5700 bases) were determined by Northern blot analysis using a specific probe and are indicated by the bar labeled al(l1). As a low molecular weight RNA marker, functional mRNA encoding the small cartilage proteoglycan core protein (43 kDa), as determined by cell-free translation/immunoprecipitation, peaks a t the expected size between fractions 24 and 34 (1200-2900 bases) in the bovine gradient. Fractions were pooled as indicated. The link protein precursor was immunoprecipitated from cell-free translation products of pooled RCS ( b ) or BAC (c) RNA fractions. Panel A (in b and c) shows translation products and artifacts of the reticulocyte lysate translation system with no exogenous mRNA (lane C), trans- lation products of unfractionated RNA (lane T), and translation products of pooled RNA fractions (lanes 1-6). From lanes 1-6, trans- lation products are progressively enriched in high molecular weight

evident (as indicated by the bar labeled aI(I1)) in gradient fractions over the correct size range (Sandell, 1984) for human type I1 collagen mRNA (5700 bases). Fractions were pooled as indicated (Fig. 5a) for translation studies.

Results of immunoprecipitation from cell-free translation products of gradient-fractionated RNA are shown in Fig. 56 for RCS RNA and in Fig. 5c for BAC RNA. From the top to the bottom of the gradient there is an enrichment in mRNA encoding protein translation products of progressively higher molecular weight (panel A in Fig. 5, b and c, lanes 1-6). The prominent bands in the control translation (panel A in Fig. 5, b and c, lane C) to which no exogenous mRNA was added are present in all translation mixtures and represent artifacts of the reticulocyte lysate translation system. All of the bands in the translation products of fractionated mRNA are present in the lanes showing the translation products of unfraction- ated mRNA (panel A in Fig. 5, b and c, lane T).

Panel B in Fig. 5, b and c, shows immunoprecipitates of the same translation products of pooled RNA fractions. In panel B of Fig. 56 (RCS), a band with an apparent molecular mass of 42 kDa is seen in lanes 3-6 and is most intense in lane 5. Similarly, in panel B of Fig. 5c (BAC), a 42-kDa band is seen in lanes 3-6 and is most intense in lane 4. Translatable RCS link protein mRNA is therefore estimated to be 4000 to 5500 bases, and BAC link protein mRNA is estimated to be 3000 to 4500 bases in length, as compared to the distribution of 18 and 28 S rRNA across the gradient.

Swarm Rat Chondrosarcoma Link Protein Biosynthesis- RCS precursor LP is an apparently single protein approxi- mately 42 kDa in size. This is not unexpected since the RCS tumor (Oegema et al., 1975, 1977) and RCS cells in culture (Kimura et al., 1980) produce a single link protein. The LP produced by cultured RCS cells migrates on SDS-PAGE slightly slower than the unmodified translation products. The link protein precursor is approximately 2.5 kDa smaller in apparent molecular mass than the glycosylated link protein which has been shown to contain a single Asn-linked carbo- hydrate chain located at amino acid 41 of the mature protein

R. S. Sawhney and L. J. Sandell, manuscript in preparation.

proteins. Panel B (in b and c) shows translation products of unfrac- tionated RNA (lane T) and immunoprecipitates of translation prod- ucts of pooled RNA fractions. The 42-kDa RCS link precursor is most abundant in lane 5 (b) , indicating that pooled RNA fractions 41-50 (4000-5500 bases) are enriched in translatable link protein mRNA. The 42-kDa BAC link precursor is most abundant in lane 4 (c), indicating that BAC RNA fractions 35-44 (3000-4500 bases) are enriched in translatable link protein mRNA.

Biosynthesis and Cell-free Translation of Link Protein 1035

(Neame et al., 1986). When addition of this Asn-linked car- bohydrate chain is inhibited by tunicamycin, a single smaller protein band is generated (Lohmander et al., 1983). The similarity in size between the LP2 produced by RCS cells and the primary gene product (Fig. 1) makes it unlikely that link protein is synthesized in a large “pro” form that is proteolyt- ically processed to generate the functional protein, as occurs with other extracellular matrix proteins such as the intersti- tial collagens. The ability to translate the LP precursor in vitro will allow future functional analyses of early biosynthetic forms of the link protein.

Bovine Cartilage Link Protein Biosynthesis-In contrast to RCS cells, the biosynthesis of bovine articular cartilage link proteins appears to be more complex. Bovine articular carti- lage chondrocytes in culture produce link proteins that are similar to those isolated from articular cartilage extracellular matrix (Treadwell et al., 1980a) and can consequently be used as a model system for their biosynthesis. When mRNA from these cultures is translated, an apparently single link protein product can be immunoprecipitated using an antiserum which precipitates multiple link proteins from bovine chondrocytes. The reduction of multiple forms of LP to a single smaller form after treatment of the cultures with tunicamycin is consistent with our results from cell-free translation of bovine chondrocyte mRNA. The LP3-sized protein, which we pro- pose to designate LP3-tu to indicate its unique origin, may be the precursor of the two larger link proteins. Taken together, our data suggest that a single protein is produced which gives rise to multiple mature forms. In a previous report of cell-free translation of bovine cartilage chondrocyte mRNA (Treadwell et al. 1980b) two proteins (M, = 41,000 and 28,000) were immunoprecipitated with antiserum specific for link protein. The 41.0-kDa product may be identical to the 42.0-kDa L P precursor that we have observed, whereas the smaller protein may be a degradation product.

In cartilage, three forms of link protein have been isolated (M, = 42,000-48,000). This heterogeneity in link protein forms, usually with LP1 and LP2 predominant, has been observed in bovine nasal cartilage (Tang et al., 1979; Baker and Caterson, 1979), canine articular cartilage (Fife et al., 1985), chicken xyphoid cartilage (McKeown-Longo et al., 1982), and human articular cartilage (Roughley et al., 1982). In older human cartilage, even more species of link protein have been isolated, possibly resulting from age-related limited proteolytic cleavage in vivo (Mort et al., 1983). Preliminary results of cell-free translation/immunoprecipitation using hu- man chondrocyte RNA suggests that a single link protein precursor is produced which is similar in size to the bovine and rat precursor.6

The difference between bovine LP1 and LP2 has been shown to reside mainly in content and type of carbohydrate (Baker and Caterson, 1979). Since link protein heterogeneity involves differences in constituent oligosaccharides, it may actually originate in the number of potential Asn-linked car- bohydrate attachment sites in the primary amino acid se- quence of the molecule. The single RCS link protein is known to contain a single site for addition of an Asn-linked carbo- hydrate chain (Neame et al., 1986). Multiple link proteins have been purified from chicken cartilage (McKeown-Longo et al., 1982). Deak et al. (1986) have isolated cDNA clones for chicken link protein in which the deduced amino acid se- quence indicates two potential sites for N-glycosylation. In tissues containing multiple link proteins, heterogeneity may therefore arise by glycosylation a t one or multiple sites on a

T. M. Hering, M. B. Goldring, and L. J. Sandell, unpublished data.

single precursor. Alternatively, multiple precursors may exist containing different numbers of potential sites for N-linked carbohydrate attachment. Although Doege et al. (1986) have shown evidence to indicate that there is only a single link protein gene, multiple precursors could arise through an al- ternate splicing mechanism which operates to produce a single precursor in RCS cells and multiple precursors in other cell types expressing multiple link proteins. At present, the num- ber of sites available for Asn-linked oligosaccharide addition in bovine link protein is unknown. This question will be resolved by analysis of bovine link protein clones, currently in progress in our laboratory.

We have observed that LP3 accumulates in overnight la- beled bovine chondrocyte culture medium, and is present to a smaller extent in immunoprecipitates from the cell layer, where it is apparently associated with the extracellular matrix. LP3 is not a cellular product, since it is not seen in immuno- precipitates from pulse-labeled chondrocyte cell lysates under the same culture conditions. These results indicate that LPs smaller then LP2 isolated from the medium of cell culture or the matrix of tissue are likely to be degradation products generated by protein or carbohydrate chain clipping. We propose the designation of this molecular species as LP3-ec to indicate its extracellular origin and to distinguish it from other LP3-sized molecules. I t is not known at this time whether LP3-tu and LP3-ec are identical. Previous data have indicated that link proteins migrating in the region of LP3 can also be generated by trypsin (Faltz et al., 1979), papain (Hascall and Heinegard, 1974), or clostripain (Caputo et al., 1980) digestion of LP1 and LP2. We propose the designation of these species of LP as LP3-pd to indicate their origin from proteolytic degradation.

Link Protein mRNA-The greatest abundance of the link protein precursor is obtained from translation products of mRNA which is much larger than the theoretical minimum size (1000 bases). Translatable RCS link protein mRNA ap- pears to be within the range of 4000-5500 bases and translat- able BAC link protein mRNA in the 3000-4500 base range. Three mRNAs, 2.5, 5.8, and 6.0 kb have been detected in chicken sternal mRNA by hybridization with a cloned cDNA coding for link protein (Deak et al., 1986). In the cloned chicken sternal cDNA, the sequences encoding link protein are located at the 5’ end of the mRNA and cover approxi- mately 1000 nucleotides. The remaining sequence at the 3’ end is presumably untranslated. The sequence of the chick mRNAs larger than 2.5 kb is not known at this time and the role played by the nontranslated region of these mRNAs is unclear. However, mRNAs coding for myelin basic proteins (de Ferra et al., 1985) are much larger than required to code for their protein products due to the presence of a tract of untranslated sequence at the 3‘ end which is three to five times longer than the coding region.

While this work was in progress, a link protein clone was isolated from a rat chondrosarcoma cDNA library (Doege et al., 1985, 1986). DNA isolated from this clone hybridized to four mRNAs ranging in size from 1.5 to 5.5 kb on Northern blots of total rat chondrosarcoma RNA. The smaller mRNAs appear to be present in a concentration similar to the larger mRNAs. Our finding that translatable RCS mRNA is most abundant in a 4.0-5.5 kb size range suggests that not all of the rat chondrosarcoma mRNAs detectable by Northern blot hybridizations can be translated into link protein in vitro. This apparently differential usage of mRNA species may play a role in the regulation of link protein synthesis.

Acknowledgments-We would like to thank Dr. James Kimura for RCS cells, Dr. Ira Wool for nucleic acid probes, Dr. Bruce Caterson

1036 Biosynthesis and Cell-free Translation of Link Protein

and Dr. James Christner for antisera, Edward Dudek for expert technical assistance, Dr. Rajinder Sawhney for helpful discussions, and Dr. William Upholt for a critical reading of the manuscript.

REFERENCES Alwine, J. C., Kemp, D. J., Parker, B. A., Reiser, J., Renart, J., Stark,

Baker, J. R., and Caterson, B. (1979) J. Bwl. Chem. 254,2387-2393 Baker, J. R., Caterson, B., and Christner, J. E. (1983) in Limb

Development and Regeneration (Kelley, R. O., Goetinck, P. F., and MacCabe, J. A., eds) Part B, pp. 17-24, Alan R. Liss, Inc., New York

Bonner, W. M., and Laskey, R. A. (1974) Eur. J. Biochem., 4 6 , 83- 90

Bonnet, F., Perin, J.-P., and Jolles, P. (1978) Biochim. Biophys. Acta 5 3 2 , 242-248

Buckwalter, J. A., Rosenberg, L. C., and Tang, L.-H. (1984) J. Biol.

Caterson, B., Baker, J. R., Christner, J. E., Lee, Y., and Lentz, M. (1985) J. Biol. Chem. 260, 11348-11356

Caputo, C. B., MacCallum, D. K., Kimura, J. H., Schrode, J., and Hascall, V. C. (1980) Arch. Biochem. Biophys. 204,220-233

Chandrasekhar, S., Kleinman, H. K., and Hassell, J. R. (1983) J. Biol. Chem. 268,6226-6231

Choi, H. U., Tang, L. H., Johnson, T. L., and Rosenberg, L. (1985) J. Biol. Chem. 260, 13370-13376

Deak, F., Kiss, I., Sparks, K. J., Argraves, W. S., Hampikian, G., and Goetinck, P. F. (1986) Proc. Natl. Acad. Sei., U. S. A. 83, 3766- 3770

de Ferra, F., Engh, H., Hudson, L., Kamholz, J., Puckett, C., Moli- neaux, S., and Lazzarini, R. A. (1985) Cell 43, 721-727

Doege, K. J., Hassell, J. R., Caterson, B., and Yamada, Y. (1985) J. Cell. Biol. 101, 7a (abstr.)

Doege, K. J., Hassell, J . R., Caterson, B., and Yamada, Y. (1986) Proc. Natl. Acud. Sci., U. S. A. 83, 3761-3765

Faltz. L. L., Caputo, C. B., Kimura, J. H., Schrode, J., and Hascall, V. C. (1979) J. Biol. Chern. 254, 1381-1387

Fife, R. S., Caterson, B., and Myers, S. L. (1985) J. Cell Biol. 100, 1050-1055

Gardell, S., Baker, J. R., Caterson, B., Heinegard, D., and Roden, L. (1980) Biochem. Biophys. Res. Commun. 95, 1823-1831

Hascall, V. C., and Heinegird, D. (1974) J. Biol. Chem. 249,4232- 4241

G. R., and Wahl, G. M. (1979) Methods Enzymol. 68, 220-242

Chem. 259,5361-5363

(1979) J. Bi01. Chem. 254 , 2600-2609 Kimura, J. H., Hardingham, T. E., and Hascall, V. C. (1980) J. Biol.

Kimura, J. H., Shinomura, T., and Thonar, E. J.-M. A. (1987)

Kuettner, K. E., Pauli, B. U., Gall, G., Memoli, V. A., and Schenk,

Laemmli, U. K. (1970) Nature 227,680-685 Lohmander, L. S., Fellini, S. A., Kimura, J. H., Stevens, R. L., and

Hascall, V. C. (1983) J. Biol. Chem. 268, 12280-12286 Maniatis, T., Fritsch. E. F., Sambrook, J. (1982) Molecuhr Cloning,

A Laboratory Manual, pp. 202-203, Cold Spring Harbor Laboratory, Cold Spring Harbor, New York

McKeown-Longo, P. J., Sparks, K. J., and Goetinck, P. F. (1982) Collagen Relat. Res. 2,231-244

Mort. J. S., Poole, A. R., and Roughley, P. J. (1983) Biochem. J. 214 ,

Neame, P. J., Christner, S. E., and Baker, S. R. (1986) J. Biol. Chem.

Oegema, T. R., Jr., Hascall, V. C., and Dziewiatkowski, D. D. (1975)

Oegema, T. R., Jr., Brown, M., and Dziewiatkowski, D. D. (1977) J.

Pelham, H. R. B., and Jackson, R. J. (1976) Eur. J. Bwchem. 67,

Roughley, P. J., Poole, A. R., and Mort, J. S. (1982) J. Biol. Chem.

Sandell, L. J., (1984) J. Celt Biol. 9 9 , 4095 (abstr.) Sandell, L. J., and Daniel, J. C. (1987) Connect. Tissue Res., in press Strohman, R. C., Moss, P. S., Micou-Eastwood, J., Spector, D.,

Przybyla, A., and Paterson B. (1977) Cell 10 , 265-273 Struck, D. K., and Lennarz, W. J. (1980) in The Biochemistry of

Glycoproteins nnd Proteoglycans (Lennarz, W. J., ed) pp. 35-83, Plenum Press, New York

Takatsuki, A., and Tamura, G. (1971) J. Antibiot. (Tokyo) 24, 232- 238

Tang, L.-H., Rosenberg, L., Reiner, A., and Poole, A. R. (1979) J. Bwl. Chem. 254,10523-10531

Thomas, P. S. (1980) Proc. Natl. Acud. Sci. U. S. A. 77,5201-5205 Treadwell, B. V., Shader, L., Towle, C. A., Mankin, D. P., and

Mankin, H. J. (1980a) Biochem. Biophys. Res. Commun. 9 4 , 159- 166

Treadwell, B. V., Mankin, D. P., Ho, P. K., and Mankin, H. J. (1980b) Biochemistry 19,2269-2275

Chem. 255.7134-7143

Methods Enzymol. 144,372-393

R. K. (1982) J. Cell Biol. 93,743-750

269-272

261,3519-3535

J. Biol. Chem. 250,6151-6159

Biol. Chem. 252,6470-6477

247-256

257 , 11908-11914

Hardingham, T. E. (1979) Biochem. J. 177,237-247 Vijayagopal, P., Radhakrishnamurthy, B., Srinivasan, S. R., and Kimura, J. H., Hardingham, T. E., Hascall, V. C., and Solursh, M. Berenson, G. S. (1985) Biochin. Biophys. Acta 8 3 9 , 110-118