Identification of a Monoclonal Antibody That Specifically Recognizes Corneal and Skeletal Keratan Sulfate MONOCLONAL ANTIBODIES TO CARTILAGE PROTEOGLYCAN*

Monoclonal antibodies were raised against proteo- glycan core protein isolated after chondroitinase ABC digestion of human articular cartilage proteoglycan monomer. Characterization of one of the monoclonal antibodies (1/20/5-D-4) indicated that it specifically recognized an antigenic determinant in the polysaccha- ride structure of both corneal and skeletal keratan sulfate. Enzyme immunoassay analyses indicated that the mouse monoclonal IgGl recognized keratan sulfate in native proteoglycan aggregate and proteoglycan monomer preparations isolated from hyaline cartilages of a wide variety of animal species (human, monkey, cow, sheep, chicken, and shark cartilage). The 1/20/5- D-4 monoclonal antibody did not recognize antigenic determinants on proteoglycan isolated from Swarm rat chondrosarcoma. This finding is consistent with several biochemical analyses showing the absence of keratan sulfate in proteoglycan synthesised by this tissue. A variety of substructures isolated after selective cleavage of bovine nasal cartilage proteoglycan (Hei-negHrd, D., and Axelsson, J. (1977) J. Biol. Chem. 252, 1971-1979) were used as competing antigens in radioimmunoassays to characterize the specificity of the 112015-D-4 immunoglobulin. Substructures de- rived 1/20/5-D-4 monoclone immunoglobulin containing 7-1 heavy chains chains. The of 1/20/5-D-4 an monoclone facilitated its use in radioimmunoassay procedures utilizing S. aureus protein A (42). Radioimmunoassay for Detecting the Antigen Specificity of the I / 20/5-D-4 Monoclonal Immunoglobulin-The RIA procedure used was a modification of one described in previous publications (21, 24, 41). A higher pH was used in the buffers to facilitate the binding of mouse immunoglobulin to the heat- and formalin-treated S. nureus cells, which allows all subclasses of immunoglobulin except IgM to bind to the S. aureus (42). Incubations were performed in a solution of 1% bovine serum albumin, 0.5% deoxycholate, 0.25% Nonidet P-40, 0.02% NaN3 in 0.15 M sodium phosphate, pH 8.1. The presence of detergents in the incubation and wash buffers ensured the complete solubilization of antigens. Antigen-antibody complexes bound to S. aureus pellets were'washed with 0.15 M sodium phosphate, pH 8.1. Antigens (HAC-PG-Core(ABC), BNC-PG-Core(ABC), and RC- PG-Core(ABC)) were iodinated with '%I using chloramine-T (43). Competitive binding radioimmunoassays used the conditions de- scribed in previous publications (21, 24, 41), except for the buffer changes indicated above, and measured the ability of unlabeled BNC proteoglycan antigens to compete with '251-BNC-PG-Core(ABC) for a known dilution of the 1/20/5-D-4 culture medium. The per cent inhibition given by each unlabeled antigen is expressed as 100 - (100 X cpm bound in the presence of unlabeled antigen/cpm bound in the absence of unlabeled antigen).

Monoclonal antibodies were raised against proteoglycan core protein isolated after chondroitinase ABC digestion of human articular cartilage proteoglycan monomer. Characterization of one of the monoclonal antibodies (1/20/5-D-4) indicated that it specifically recognized an antigenic determinant in the polysaccharide structure of both corneal and skeletal keratan sulfate. Enzyme immunoassay analyses indicated that the mouse monoclonal IgGl recognized keratan sulfate in native proteoglycan aggregate and proteoglycan monomer preparations isolated from hyaline cartilages of a wide variety of animal species (human, monkey, cow, sheep, chicken, and shark cartilage). The 1/20/5-D-4 monoclonal antibody did not recognize antigenic determinants on proteoglycan isolated from Swarm rat chondrosarcoma. This finding is consistent with several biochemical analyses showing the absence of keratan sulfate in proteoglycan synthesised by this tissue.
A variety of substructures isolated after selective cleavage of bovine nasal cartilage proteoglycan (Hei-negHrd, D., and Axelsson, J. (1977) J. Biol. Chem. 252,[1971][1972][1973][1974][1975][1976][1977][1978][1979] were used as competing antigens in radioimmunoassays to characterize the specificity of the 112015-D-4 immunoglobulin. Substructures derived from the keratan sulfate attachment region of the proteoglycan (keratan sulfate peptides) showed the strongest inhibition. Both corneal and skeletal keratan sulfate peptides as competing antigens in radioimmunoassays showed similar inhibition when compared on the basis of their glucosamine content. Therefore, the 1/20/5-D-4 monoclonal antibody appears to recognize a common determinant in their polysaccharide moieties. Chemical desulfation of the keratan sulfate reduced the antigenicity of the glycosaminoglycan. The antibody did not recognize determinants present in dermatan sulfate, heparin, heparin sulfate, or hyaluronic acid.
Proteoglycans are ubiquitous components of connective tissues. They are particularly abundant in tissue where there is greatest development of extracellular matrix material (e.g. cartilage, skin, bone, and teeth). They endow the extracellular matrix, and the tissue, with many of its characteristic bio-* This work was funded by National Institutes of Health Grants AM27791, AM32474, AM27308, and HL28195. 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. mechanical and biophysical properties (1). Proteoglycans from bovine nasal cartilage and bovine tracheal cartilage have been the most thoroughly studied (2,3) although in recent years studies on the Swarm rat chondrosarcoma have provided considerable information on the structure and biosynthesis of its constituent cartilage proteoglycans (4)(5)(6)(7).
In cartilage, proteoglycans occur as aggregates (2,3,8,9) which consist of three classes of components: proteoglycan monomers, hyaluronic acid, and link proteins (2,3,10,11). A model for the cartilage proteoglycan aggregate has been proposed (12,13). In this model, the proteoglycan monomers are pictured as asymmetric molecules. A region of the proteoglycan monomer core protein, the hyaluronic acid-binding region, specifically associates with a central strand of hyaluronic acid (2,3). Beyond the hyaluronic acid-binding region of the proteoglycan monomer, the protein core bears a region containing predominantly keratan sulfate side chains and beyond that is the chondroitin sulfate attachment region of the molecule where a few keratan sulfate chains are found interspersed with the relatively predominant chondroitin sulfate chains. In addition, Nand 0-linked oligosaccharide structures distributed on the polypeptide backbone of the proteoglycan monomers have been recently described (6,14,15). The link proteins of the proteoglycan aggregate stabilize the association of proteoglycan monomers with hyaluronic acid ( 2 , 3, 16). This stabilization occurs through specific interactions of the link protein with both the proteoglycan monomer and the hyaluronic acid (17,18).
Determination of proteoglycan structure by conventional biochemical procedures has been compIicated by the polydispersity and heterogeneity which appears common to all classes of connective tissue proteoglycan. Potentially, immunological methods may be particularly useful, as antibodies to proteoglycan may be used to determine specific antigenic determinants, regardless of polydispersity and heterogeneity of the proteoglycan. Substructures of cartilage proteoglycan monomer (2,19) have distinct biochemical characteristics and can therefore, potentially, be recognized and quantitated by specific immunological means. Polyclonal antibodies recognizing link protein and several of the proteoglycan monomer substructures have been reported (20)(21)(22)(23)(24) and have been used to quantitate and immunohistochemically localize proteoglycans in connective tissue preparations.
Monoclonal antibodies, raised by a technique based on that of Kohler and Milstein (25), recognize determinants on hyaluronidase-treated proteoglycan monomer from chick sternal cartilage (26, 27). One of these monoclonal antibodies has been used to isolate proteoglycan core synthesized in cell-free systems (26), and the other was shown to specifically recognize saturated oligosac-8848 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from charides of chondroitin 6-sulfate attached to the protein core of hyaluronidase-treated proteoglycans (27). Recently, we have described the production and partial characterization of several monoclonal antibodies to link protein and to chondroitinase ABC-treated proteoglycan monomer from bovine nasal cartilage (28). The immunoglobulins produced by the hybridomas were specific for many different determinants of proteoglycan monomer and link protein. In this paper, we describe the characterization of a monoclonal antibody that specifically recognizes a determinant in the characteristic polysaccharide of keratan sulfate. This monoclonal antibody can be used in radioimmunoassay procedures to specifically detect and quantitate keratan sulfate in proteoglycan preparations.

EXPERIMENTAL PROCEDURES
Materials-The hydrochloric acid used for acid hydrolysis was Aristar grade (British Drug Houses). All other reagents used were of analytical reagent grade. Chondroitinase ABC was purchased from Miles Laboratories. Porcine stomach mucin (type II), bovine serum albumin type V, guanidine hydrochloride, benzamidine, phenylmethylsulfonyl fluoride, EDTA, and 2-mercaptoethanol were obtained from Sigma. Nonidet P-40 was obtained from Gallard-Schlesinger and 6-aminohexanoic acid was purchased from Eastman Kodak. Freund's complete adjuvant and incomplete adjuvant were bought from Grand Island Biological Company and lZ5I was purchased from Amersham. Tissue culture dishes (100 mm) were from Corning (catalog No. 20520). All other tissue culture dishes (6-, 24-, and 96-we11 culture dishes) were purchased from Costar. Medium (RPMI 1640), sterile Dulbecco's phosphate-buffered saline, media additives (fetal calf serum, glutamine, Pen-Strep, and Fungisone), Lindbro EIA' plates and the Titertek Multiskan for reading the EIA plates were purchased from Flow Laboratories. Gels for gel chromatography were obtained from Pharmacia Fine Chemicals. CPG 2500 was a generous gift from Dr. Paul Goetinck (University of Connecticut, Storrs, CN) and was originally purchased from Electronucleonics. CPG 200-Glycophase was obtained from Pierce. Formalin-treated, heat-inactivated Staphylococcus aureus was prepared by published procedures (29). Dermatan sulfate, heparin, and heparan sulfate were generous gifts from Dr. Richard Reynertson and Dr. Lennert Roden, University of Alabama in Birmingham, and were prepared by published procedures (30). Human and dog blood group glycolipids, prepared by published procedures (31) Column Chromatography-Preparative columns of Sepharose CL-8.0. The occurrence of proteoglycan, peptides, and proteoglycan fragments in the column effluents was monitored by absorbance at 206 nm on a UVicord (LKB Instruments) or at 214 nm on a UV I1 (Pharmacia Fine Chemicals) ultraviolet monitor, respectively. Fractionated materials were recovered from pooled fractions by lyophilization.
Analytical Methods-Uronic acid determinations were according to the carbazole method of Bitter and Muir (32). Amino acid analyses and hexosamine analyses were performed after 20-and 7-h hydrolyses, respectively, in 6 M HCI a t 105 "C and quantitated on a Beckman 119CL automatic amino acid analyzer (33). Norleucine was used as an internal standard in both analyses. Protein (50 pg) or glycosaminoglycan (20 pg) was hydrolyzed in Reacti-vials (Pierce Chemical Co.) under nitrogen (total hydrolysis volume, 400 p l ) . Sulfate analyses were performed using a modified Rhodizonate procedure described by Silvestri et al. (34).
The proteoglycan aggregate fraction (A1)' and proteoglycan monomer fraction (AlD1) were obtained after CsCl equilibrium density gradient centrifugation under "associative" and "dissociative" conditions, respectively. The proteoglycan fractions were dialyzed at 4 "C for 24 h against deionized water, 0.1 M NaCI, and two further changes of deionized water prior to their lyophilization and storage in a desiccator a t room temperature. Stock solutions of proteoglycan A1 (2 mg/ ml) and AlDl (4 mg/ml) were prepared in PBS-azide and stored a t 4 "C for analyses in the EIA and RIA procedures described below.
Isolation of Human Articular Cartilage Proteoglycan Core Protein Used As the Antigen for Immunization of BALB/c Mice-A pooled sample of human articular cartilage from the femoral and tibial chondyles of six normal males (age range, 50-77 years) was used for extraction of human articular cartilage proteoglycan. In all cases, the cartilage was obtained at autopsy within 4 h of death, dissected from the articular surfaces, and stored a t -20 "C until required in this study. The cartilage was diced into small pieces using a scalpel on an ice-cooled dissection platform. The proteoglycan aggregate and monomer preparations were obtained by the procedures described above. The galactosamine to glucosamine ratios of aggregate and monomer preparations were 5.4 and 7.5, respectively, suggesting a high proportion of keratan sulfate (cf. proteoglycans from bovine nasal cartilage) in these preparations.
Human articular cartilage proteoglycan core protein (HAC-PG-Core(ABC)) was obtained after digestion of the proteoglycan monomer, AlDl fraction, with chondroitinase ABC. The core protein was separated from chondroitinase and unsaturated disaccharides by chromatography on Sepharose CL-GB, using conditions described previously (24). The void volume material was pooled, dialyzed exhaustively against deionized water, lyophilized, and stored desiccated at room temperature. A stock solution (800 pg/ml in sterile Dulbecco's phosphate-buffered saline, 5 X 0.5-ml aliquots) was prepared, stored at -20 "C, and used in the injection protocol described below.
Immunization, Fusion, and Cloning-Three 4-6-week-old female BALB/c mice were immunized with HAC-PG-Core(ABC) and hybridoma fusion performed according to the protocol described by Kearney and co-workers (37,38). The antigen (HAC-PG-Core(ABC)) was administered to each animal at six injection sites (hind foot pads, lateral thoracic and inguinal regions) on days 1, 3, 6,9, and 12. The antigen (0.5 ml of HAC-PG-Core(ABC), 800 pg/ml) was mixed with 0.5 ml of Freund's complete adjuvant (day l), 0.5 ml of Freund's incomplete adjuvant (day 3), and 0.5 ml of Dulbecco's PBS at days 6, 9, and 12, respectively, and 0.05 ml of the antigen mix was injected at each of the six injection sites of each mouse. Two days after the final injection (day 14), the draining lymph nodes from the regions nearest the sites of injection (axillary, brachial, inguinal, and popliteal lymph nodes) were removed aseptically under a laminar flow hood and placed in a 25-mm Petri dish containing 2 ml of Dulbecco's PBS, cooled at 4 "C in an ice bath. Lymphocytes were dispersed in the solution by teasing the lymph nodes with forceps and finally by repeated suction and aspiration through an 18-gauge needle using a 3-ml syringe. The lymphocytes were separated from adhering connective tissue by passing the cell suspensions through a sterile glass wool filter plug in a Pasteur pipette and the cells were washed twice with 40 ml of RPMI 1640 medium at 4 "C.
The characteristics of the hypoxanthine-, aminopterin-, thymidine-sensitive mouse myeloma cell line (X 63 -Ag8.653) have been described elsewhere (37,39). This cell line is available from the American Type Culture Collection. The mouse myeloma cell line (X 63 -Ag8.653) was cultured in normal culture media (RPMI 1640 medium containing 15% fetal calf serum, 0.1 M NaHC03, 0.01 unit/ ml PenStrep, 0.001 unit/ml Fungisone, and 0.001 M 2-mercaptoethanol) in 100-mm tissue culture Petri dishes (Corning, 20520). Myeloma cells from eight Petri dishes growing at log phase (2-5 x lo8 cells total) were washed twice with RPMI 1640 medium at 4 "C to remove fetal calf serum and medium components. Lymphocytes (-1 The abbreviations A1 and AlDl used to describe proteoglycan fractions follow the notation suggested by Heinegird (36) for proteoglycan aggregate and monomer, respectively. by guest on March 24, 2020 http://www.jbc.org/ Downloaded from X 10') cells and the myeloma cells were suspended in 10 ml of RPMI 1640 medium, counted on a Coulter counter, and mixed at a lymphocyte to myeloma ratio of 2:l. The cells were pelleted by centrifugation, the supernatant was removed, and 1.5 ml of 20% polyethylene glycol 4000 (Fisher Scientific), previously heated to 37 "C, was added dropwise with vigorous shaking. After the addition of the polyethylene glycol (1-2 min), 15 ml of RPMI medium (at 37 "C) was slowly added dropwise with gentle agitation. The fusion products were washed once with 40 ml of RPMI 1640 medium at room temperature, pelleted by centrifugation, and resuspended in 200 ml of culture media containing 100 pM hypoxanthine, 0.4 phi aminopterin, 1.6 p~ thymidine and feeder cells from a 3-ml peritoneal lavage of a normal BALB/c mouse. The cell suspension was aliquoted (-1 ml/well) into eight 24-well Costar plates and cultured at 37 "C in an incubator in the presence of 5% COS. After 10 days of tissue culture in hypoxanthine-aminopterin-thymidine medium, a 0.5-ml aliquot of each culture well was taken and tested in an EIA for the presence of mouse immunoglobulins directed against HAC-PG-Core(ABC). Complete culture medium (0.5 ml) was added to all wells and 4 days later, the media were retested for antibody production.
antibodies which recognized HAC-PG-Core(ABC) were identified.
After 2 weeks of culture, several strongly positive clones producing Monoclonal cell lines were obtained by the following procedure. After the cells in an antibody-positive well of the 24-well plate had grown to confluency, 2-pl aliquots of the cell suspension were diluted in 20 ml of complete culture medium containing 3 ml of feeder cells. The diluted cells were aliquoted into 96-well Costar plates (200 pllwell). After 2 weeks, the media from each of the wells were tested by EIA for reactivity with HAC-PG-Core(ABC). If antibody-positive cell colonies appeared in less than one-third of the wells, it was found by visual inspection of colony morphology that essentially all of the colonies arose from a single cell. Profusely growing clones were subcloned as described above and the cells were aliquoted for storage under liquid nitrogen or grown up in larger tissue culture dishes. The culture supernatants from these dishes were used in the EIA and RIA described below. Hybridoma cells from confluent culture dishes were subsequently injected intraperitoneally into female retired breeder BALB/c mice injected with 0.5 ml Pristane (2,6,10,14-tetramethylpentadecane, Aldrich Chemical Co.) and the ascites fluid was recovered 2 weeks later for large scale purification of the monoclonal antibodies. Preparation of BNC Proteoglycan Substructures for Use As Antigens in Determining the Specificity of the 1/20/5-D-4 Hybridoma Immunoglobulin-BNC proteoglycan monomer substructures were used for characterization of the 1/20/5-D-4 monoclonal antibody specificity.
The procedures for isolating the BNC proteoglycan substructures are depicted in Fig. 1. BNC-PG-Core(ABC) (Fig. L4) was prepared after chondroitinase ABC digestion of BNC-A1D1 (24). BNC chondroitin sulfate-oligosaccharide peptides (Fig. 1, F and H) and BNC keratan sulfate peptides (Fig. 1, E and G) were prepared after protease digestion of the BNC-PG-Core(ABC) according to the methods described by Heinegird and Axelsson (19). A minor modification of this procedure was the use of CPG 200 column chromatography instead of Sepharose CL-GB chromatography which gave a more rapid separation of the keratan sulfate peptides from the chondroitin sulfate oligosaccharide peptides. BNC chondroitin sulfate-oligosaccharide peptides and BNC keratan sulfate peptides samples (Fig. 1) were obtained by pooling appropriate fractions after gel filtration and concentrated by lyophilization. Freeze-dried samples were redissolved in 0.15 M NaCl prior to amino acid and hexosamine analysis or use in the competitive binding RIA described below.
BNC glycosaminoglycan fractions (Fig. 1, B, C, and D) were obtained from BNC-A1D1 after digestion with trypsin and fractionation on DEAE-cellulose using published procedures (30). Three fractions (Fig. 1, antigens B, C, and D) from the digest were obtained by stepwise elution at 0.3, 0.4, and 0.5 M NaCl, respectively from DEAE-cellulose. The glycosaminoglycan fractions accounted for 15, 69, and 16% of the dry weight of the glycosaminoglycan recovered from the digest, respectively. Eluted fractions were dialyzed exhaustively against deionized water and freeze-dried. Stock solutions (10 mg/ml in PBS-azide) were prepared and used for amino acid analyses, hexosamine analyses, and EIA or RIA analyses of 1/20/5-D-4 immunoglobulin specificity. A corneal keratan sulfate glycosaminoglycan fraction was prepared from papain/pronase digests of bovine corneas according to the procedures described by McCarthy and Baker (40). Desulfated keratan sulfate was isolated after methanolic HCl treatment of the corneal keratan sulfate fraction (40). Sulfate analyses (34) of this desulfated corneal keratan sulfate fraction indicated that 75% of the ester sulfate had been removed.
Enzyme Immunoassay for Detection and Characterization of the Cloml Hybrid Specificities-The method for preparing alkaline phosphatase-labeled purified goat antibodies (goat anti-mouse K and h light chains) has been described elsewhere (37). Antibody-secreting hybrids were detected by a modification of an EIA procedure described in a recent publication (41). Antigens were added and incubated for 90 min at 37 "C. The plates were washed three times with PBS-azide. Enzyme-linked second antibody (alkaline phosphatase-conjugated goat anti-mouse K and X light chains) was added (200 pllwell) and similarly incubated. The plates were washed three times with PBSazide and then 200 pl of alkaline phosphatase substrate (p-nitrophenyl phosphate, 1 mg/ml, in 0.25 mM MgCIz, 1 M diethanolamine, pH 9.8) were added to each well and the plates were incubated 30-90 min at 37 "C until optimal color development occurred. The reaction was stopped by the addition of 5 M NaOH (50 pl/well). Clonal antibody-positive wells were identified by the presence of yellow color resulting from the conversion of the p-nitrophenyl phosphate to pnitrophenol by the enzyme-linked second antibody. The p-nitrophenol was quantitated by measuring the absorption at 405 nm of the solution from each well (EIA Multiskan, Flow Laboratories).
Mouse immunoglobulin subclass and heavy and light chain specificities were determined by a modification of the above EIA procedure. HAC-PG-Core(ABC) was coated on the EIA plate, unreacted sites were blocked with EIA buffer, and the supernatants from the wells of positive clones were incubated as described above. Alkaline phosphatase-conjugated goat anti-mouse immunoglobulin subclass, heavy and light chain-specific second antibodies (a generous gift from Dr. John F. Kearney, University of Alabama in Birmingham) were used to characterize the hybridoma monoclones. The 1/20/5-D-4 monoclone was identified as producing immunoglobulin containing 7-1 heavy chains and K light chains. The identification of 1/20/5-D-4 as an IgG1-producing monoclone facilitated its use in radioimmunoassay procedures utilizing S. aureus protein A (42).
Radioimmunoassay for Detecting the Antigen Specificity of the I / 20/5-D-4 Monoclonal Immunoglobulin-The RIA procedure used was a modification of one described in previous publications (21,24,41). Competitive binding radioimmunoassays used the conditions described in previous publications (21,24,41), except for the buffer changes indicated above, and measured the ability of unlabeled BNC proteoglycan antigens to compete with '251-BNC-PG-Core(ABC) for a known dilution of the 1/20/5-D-4 culture medium. The per cent inhibition given by each unlabeled antigen is expressed as 100 -(100 X cpm bound in the presence of unlabeled antigen/cpm bound in the absence of unlabeled antigen).

RESULTS
Preliminary examination of the specificity of the 1/20/5-D-4 monoclonal supernatants in an EIA indicated that the immunoglobulin recognized an antigenic determinant common to proteoglycan preparations from a wide variety of animal species and tissues but not to Swarm rat chondrosarcoma proteoglycan (Fig. 2). Analyses in the EIA indicated that the antibody recognized antigens from articular cartilage to a greater degree than preparations from nasal septum. As keratan sulfate is absent from rat chondrosarcoma proteoglycan and enriched in articular antigen proteoglycans, these results suggested that 1/20/5-D-4 may be recognizing some part of keratan sulfate.
EIA analyses of the 1/20/5-D-4 immunoglobulin subclass were performed and indicated that the hybridoma was synthesizing immunoglobulins of the IgGl subclass containing K light chains (as mentioned above). The results of RIA analyses using serial dilutions of the 1/20/5-D-4 monoclonal media versus '251-labeled HAC-PC-Core(ABC), BNC-PG-Core(ABC), and RC-PC-Core(ABC) are shown in Fig. 3. Greater than 80% of the '251-labeled HAC-PC-Core(ABC) and BNC-PC-Core(ABC) proteoglycans were bound by the Amino acid and hexosamine analyses of the antigen used for immunization and the BNC proteoglycan substructures used as competitive inhibitors in RIA analyses are given in Table I. The analyses (Table   1) are consistent with the identifications and descriptions of the fractions. BNC proteoglycan substructures were used as antigens in competitive binding RIA to determine the specificity of the 1/20/5-D-4 monoclonal IgGl (Fig. 4). Inhibition analyses are compared relative to the amount of total amino acids present in each antigen. Proteoglycan substructures showing the greatest inhibition were those derived from the keratan sulfate attachment region of bovine nasal cartilage proteoglycan. The most inhibitory antigens were those derived from sequential chondroitinase ABC, trypsin, and chymotrypsin digestion of BNC-AlDl and the 0.5 M NaCl glycosaminoglycans eluted from DEAE-cellulose (antigens G and D, respectively; Table I and Aspartic acid  71  75  52  49  37  63  38  40  Threonine  59  40  42  43  31  45  Serine  114  96  163  151   142  174  121  Glutamic acid  150  161  155  163  180  149  195  Proline  85  133  116  134  164  59  149  Glycine  146  93  152  138  131  186  118  Alanine  56  83  60  60  42  49  42  Valine  77  60  69  66  61  80  55  Isoleucine  37  29  34  32  28 Fig. 1 and "Experimental Procedures" for description of antigen preparation. * Molar ratios. structures in the RIA is shown on Table 11. In general, inhibition is greater for keratan sulfate (glucosamine-rich) glycosaminoglycan peptide fragments. However, the 0.3 M NaCl DEAE-cellulose glycosaminoglycan fraction which is rich in glucosamine (antigen B, galactosamine/glucosamine ratio, 8.11; Tables I and 11) was found to be a relatively poor inhibitor in the RIA.
Competitive binding RIA analyses were performed using bovine corneal keratan sulfate preparations and compared with skeletal keratan sulfate samples, to determine whether the antigenic determinant recognized by the 1/20/5-D-4 immunoglobulin occurred in the polysaccharide chain or the peptide portion of the proteoglycan fragments (Fig. 5). Inhibition analyses were compared relative to the amount of glucosamine present in each antigen. The results indicate that corneal keratan sulfate, which has a distinctly different carbohydrate linkage to peptide from skeletal keratan sulfate, gave an inhibition curve similar to that of skeletal keratan sulfate in RIA analyses (50% inhibition at 17 and 55 pmol of glucosamine, respectively). Furthermore, chemical desulfation of the corneal keratan sulfate (where 75% of the ester sulfate was removed from the polysaccharide) caused a significant reduction in the antigenicity of the glycosaminoglycan (Fig.  5, Table 111). These results indicate that the antigenic determinant recognized by the 1/20/5-D-4 monoclonal antibody resides in the polysaccharide structure common to both corneal and skeletal keratan sulfate. In addition, the presence of ester sulfate on the polysaccharide significantly enhances the antigenicity of the keratan sulfate determinant recognized by the 1/20/5-D-4 IgG1. During the chemical desulfationprocess, some methylation of the glycosaminoglycan sugars may have occurred but this was subsequently reversed by alkali treatment. 4 In addition, some cleavage occurs during the desulfation p r o~e d u r e .~ However, it is unlikely that the small amount of depolymerization would significantly influence the antige-   These fucolipids contain one N-acetyllactosamine moiety similar to that found in keratan sulfate. In addition, pig gastric mucin which contains a structure similar to that found in the repeating disaccharide portion of keratan sulfate (e.g. 6-sulfated N-acetylglucosamine-linked @( 1-3) to galactose; see Ref. 44) did not show inhibition in the RIA. These results furt.her indicate the specificity of the 1/20/5-D-4 immunoglobulin for a determinant which is characteristic of the keratan sulfate glycosaminoglycan chain.

DISCUSSION
Many oligosaccharides of known structure have been shown t,o be antigenic (45). There is such a diversity of oligosaccharide structures on mammalian glycoproteins that many are likely to be antigenic. The glycosaminoglycan chains of proteoglycans from all sources tend to have the same linear disaccharide repeat sequences and are therefore less likely to be antigenic, although antibodies specific for some feature of chondroitin 4-sulfate chains have been reported (27). Recently, antibodies to oligosaccharide moieties of chondroitinase-digested proteoglycans have been raised (24). The determinants are the protein-linked oligosaccharide "remnants" of chondroitin sulfate chains.
In this paper, the characterization ofa monoclonal antibody raised against the core proteins of human articular cartilage proteoglycan is described. It is evident that the only glycosaminoglycan which it recognizes is keratan sulfate, although keratan sulfates from cornea and from cartilage bind the antibody. Desulfation of keratan sulfate decreases binding. These characteristics are also common to a rabbit antiserum to bovine corneal proteoglycan as recently described by Conrad et al. (46). The rabbit antiserum appears to possess antibody activities to the proteoglycan core protein. In contrast, the monoclonal antibody must recognize one specific structural feature of the keratan sulfate chain. To identify this structural feature will be difficult as little is known of the arrangement of differently substituted galactosyl and N-acetylglucosaminyl residues along the keratan sulfate chains. (Comparatively, much more is known of other glycosaminoglycans (30).) Thus, either or both sugar residues may be 6sulfated and the antibody may recognize a specific pattern of sulfation on two to four consecutive sugar residues. Keratanase, an endo-@-galactosidase, will be useful for generating fragments from keratan sulfate which can be characterized and then tested for competitive binding to the antigen binding site of monoclonal antibody 1/20/5-D-4.
Monoclonal antibody 1/20/5-D-4 can recognize and bind keratan sulfate chains in their native form attached to proteoglycan. Therefore, it is likely to be a useful tool for studying localization and distribution of keratan sulfate proteoglycans in tissue sections employing immunohistological methods. Indeed, this antibody has been successfully employed by Verte15 to localize keratan sulfate in the Golgi of chondrocytes.