N-Linked Keratan Sulfate in the Aggrecan Interglobular Domain Potentiates Aggrecanase Activity*

Keratan sulfate is thought to influence the cleavage of aggrecan by metalloenzymes. We have therefore produced a recombinant substrate, substituted with keratan sulfate, suitable for the study of aggrecanolysis in vitro. Recombinant human G1-G2 was produced in primary bovine keratocytes using a vaccinia virus expression system. Following purification and digestion with specific hydrolases, fluorophore-assisted carbohydrate electrophoresis was used to confirm the presence of the monosulfated Gal-GlcNAc6S and GlcNAc6s-Gal disaccharides and the disulfated Gal6S-GlcNAc6S disaccharides of keratan sulfate. Negligible amounts of fucose or sialic acid were detected, and the level of unsulfated disaccharides was minimal. Treatment with keratanases reduced the size of the recombinant G1-G2 by ∼5 kDa on SDS-PAGE. Treatment with N-glycosidase F also reduced the size of G1-G2 by ∼5 kDa and substantially reduced G1-G2 immunoreactivity with monoclonal antibody 5-D-4, indicating that keratan sulfate on the recombinant protein is N-linked. Cleavage of G1-G2 by aggrecanase was markedly reduced when keratan sulfate chains were removed by treatment with keratanase, keratanase II, endo-β-galactosidase, or N-glycosidase F. These results indicate that modification of oligosaccharides in the aggrecan interglobular domain with keratan sulfate, most likely at asparagine residue 368, potentiates aggrecanase activity in this part of the core protein.

globular domain (IGD) 1 of ϳ150 amino acids separates G1 from G2 and is substituted with keratan sulfate chains as well as O-linked and N-linked oligosaccharides.
The IGD is highly sensitive to proteinases. Cleavage in the IGD releases the entire chondroitin sulfate and keratan sulfate-rich regions essential for the biomechanical properties of aggrecan and is therefore the most detrimental for cartilage function. In pathology, proteolysis is driven by aggrecanases and, to a lesser extent, by matrix metalloproteinases. Aggrecanase was first identified as a novel activity that cleaved the aggrecan core protein at the E 373 2 374 A bond in the IGD (1-3); the products of this cleavage were found in synovial fluids from patients with osteoarthritis, joint injury, and inflammatory joint disease (4,5). Subsequently, cartilage enzymes with aggrecanase activity were revealed as members of the ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) family (6,7), and ADAMTS5 has recently been identified as the major aggrecanase in mouse cartilage (8,9).
Proteolysis by matrix metalloproteinases at the N 341 2 342 F bond in the IGD does not appear to contribute to glycosaminoglycan loss during early phases of experimental arthritis in mice. However, matrix metalloproteinase cleavage at N 341 2 342 F correlates with late-stage cartilage damage in mouse models of arthritis (10 -12), and it may also be involved in the baseline turnover of aggrecan in vitro (13) and in vivo (14). The products of in vivo proteolysis at both the matrix metalloproteinase and the aggrecanase sites have been found in humans (4,15,17,18) and in mice with experimental arthritis (11, 19 -21).
Keratan sulfate is a glycosaminoglycan with a ␤1,3-linked backbone of the repeating disaccharide Gal␤1,4GlcNAc. The majority of the GlcNAc residues and a significant proportion of adjacent Gal residues are sulfated in the C6 position (22,23), giving rise to mono-or disulfated regions, respectively. Keratan sulfate may also be fucosylated at GlcNAc6S within monosulfated chain regions (23)(24)(25) and contain sialic acid capping at the nonreducing Gal or Gal6S (26 -28).
Bovine aggrecan contains 2-3 N-linked keratan sulfate chains and 20 or more O-linked keratan sulfate chains (29), most of which are present in the keratan sulfate-rich region between the G2 globular domain and the chondroitin sulfaterich region. Substitution in the keratan sulfate-rich region is O-linked (30), however, keratan sulfate in the IGD is attached via both O-and N-linkages (29). Studies with aggrecan from porcine and bovine cartilage show that keratan sulfate attachment in the IGD is between the matrix metalloproteinase (N 341 2 342 F) and aggrecanase (E 373 2 374 A) cleavage sites, at threonine residues Thr 352 , Thr 357 , and Thr 370 and asparagine residue Asn 368 (29,31). A number of studies have shown that cleavage at the E 373 2 374 A bond is inhibited by exogenous glycosaminoglycans (32)(33)(34) and highly sulfated polymers (35). Other studies comparing the presence or absence of endogenous keratan sulfate chains have suggested that aggrecanase cleavage in the IGD may be increased in the presence of keratan sulfate and reduced when keratan sulfate is enzymatically removed (36,37). However, recombinant substrates that lack keratan sulfate are not totally resistant to cleavage by aggrecanases (38,39). Collectively, these studies suggest that keratan sulfate in the IGD may influence aggrecanolysis.
In this study, we report generation of a recombinant G1-G2 bearing N-linked keratan sulfate, and we show that this Nlinked substitution is sufficient to potentiate aggrecanase cleavage in the IGD. To our knowledge, this is the first report of a recombinant protein substituted with keratan sulfate.

MATERIALS AND METHODS
Reagents-Collagenase II was from Worthington. Monoclonal antibody 1-C-6 (40) was from the Developmental Studies Hybridoma Bank established under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences (Iowa City, IA). 1-C-6 recognizes a reduced epitope in the G1 and keratanase-digested G2 domain (40,41) of aggrecan. DMEM/F12 was from Thermo Trace. Monoclonal antibody 5-D-4, which recognizes a highly sulfated region in keratan sulfate (40,42), was a gift from Prof. Bruce Caterson (University of Wales, Cardiff, Wales). Endo-␤-galactosidase (E␤G), keratanase, keratanase II, and chondroitinase ABC were from Seikagaku. N-Glycosidase F and Oglycosidase were from Roche Diagnostics. ␣(1-3,4) Fucosidase and monogels for analysis of depolymerized keratan sulfate chains were from Prozyme, distributed by Epitope Technologies (Melb, Australia). The analytical BioSep-Sec-S 4000 size exclusion HPLC column was from Phenomenex. The vaccinia virus encoding T7-polymerase, vTF7-3, has been described previously (43).
Generation of vG1-G2-A construct for producing vaccinia virus expressing G1-G2 was made by subcloning cDNA encoding a human G1-G2 fragment of aggrecan (38) into the pTM1 vector (44) at the SacI and XhoI cloning sites. The G1-G2 insert was repositioned in the vector so that the initiating ATG codon was part of the NcoI cloning site in the pTM1 polylinker. Forward primer Agg96-2 (5Ј-AAACACGATAATAC-CATGACCACTTTACTCTGG-3Ј), which includes a 5Ј sequence complementary to a region surrounding the NcoI site in the polylinker (italics), a 3Ј sequence complementary to the G1-G2 insert (bold), and an ATG codon (underlined) common to both, was used in a PCR with reverse primer Agg96-1 (5Ј-GAGATGGCTCTGTAATGGAA-3Ј) to amplify a 500-bp fragment. This fragment contained a 15-bp 5Ј overhang that extended into the G1-G2 insert, beyond a unique NsiI site. In a separate PCR, forward primer pTM1SL.1 (5Ј-TATAAGATACACCTG-CAAAG-3Ј) and reverse primer pTM1SL.2 (5Ј-CATGGTATTATCGT-GTTTTT-3Ј) were used to amplify a 257-bp region of the pTM1 polylinker immediately adjacent to and terminating at the ATG codon of the NcoI site. This 257-bp fragment contained a central KpnI site. Splicing by overlap extension PCR (45) was used to generate a 760-bp fragment from the 500-and 257-bp templates with the pTM1SL.1 and Agg96-1 primer pairs. The 760-bp fragment was digested with KpnI and NsiI then ligated into the KpnI and NsiI restriction sites of the pTM1-G1-G2 plasmid, in place of a 560-bp cassette. The construct was sequenced prior to generation of the recombinant vaccinia virus encoding human G1-G2 (vG1-G2) under the control of the T7 promoter, as previously described (44).
Preparation of Primary Bovine Keratocytes-Corneas were removed from the eyes of freshly slaughtered steers, rinsed in PBS containing 100 units/ml penicillin and 100 g/ml streptomycin, and then incubated in a solution of 0.25% (w/v) trypsin and 0.5 mM EDTA for 20 min at 37°C. Endothelial and epithelial cell layers were removed by gentle scraping, and after further rinsing in PBS, the tissue was cut into small pieces and digested in DMEM/F12 containing 2 mg/ml collagenase II and antibiotics. After 16 h, the digest was passed through a 70-m cell strainer to remove undigested tissue debris. The cells were washed once in medium containing 10% fetal calf serum and twice in serum-free medium and then seeded at a density of 1.11 ϫ 10 5 cells/cm 2 in DMEM/ F12 medium containing 2% fetal calf serum and antibiotics. All experiments were done with non-passaged primary cells.

Expression of [ 35 S]Methionine-labeled G1-G2 in Cell
Monolayers-Confluent cell monolayers cultured in DMEM/F12 with 2% fetal calf serum and antibiotics were infected with vTF7-3 alone (control) or co-infected with vTF7-3 and vG1-G2 for 6 h at 37°C. The media were replaced with methionine and cysteine-deficient DMEM. [ 35 S]Methionine (10 Ci/ml) was added to the wells, and the infection continued at 37°C for various times up to 4 days. Aliquots of media were precipitated overnight at Ϫ20°C with 3 volumes of ethanol containing 100 mM ammonium acetate, and the precipitate was recovered by centrifugation. The pelleted, radiolabeled samples were analyzed by non-reducing SDS-PAGE and fluorography.
Large Scale G1-G2 Preparation-Preparative amounts of keratan sulfate-containing G1-G2 were expressed in primary bovine keratocytes by co-infecting cells with virus at 5 plaque-forming units/cell and, after 6 h, replacing the media with fresh media (DMEM/F12 with 2% fetal calf serum and antibiotics) as described previously (44). After 4 days, the media were collected and applied to a column of hyaluronan (HA)coupled Sepharose (46) equilibrated in PBS. Proteins bound non-specifically to HA-Sepharose were eluted with 0.5 and 1.0 M NaCl. G1-G2 bound to HA was then eluted with 4 M GuHCl, 50 mM sodium acetate, pH 5.8, and concentrated in 30-kDa cutoff Millipore spin columns. The sample was applied directly to a BioSep-Sec-S 4000 size exclusion HPLC column and eluted under dissociative conditions into tubes containing proteinase inhibitors. After desalting on Millipore spin columns, fractions containing G1-G2 were identified by Western blotting with antibodies 1-C-6 and 5-D-4 (41). The protein concentration in pooled fractions was estimated by absorbance at 278 nm, and the concentration of keratan sulfate was estimated by the 1,9-dimethylmethylene blue dye binding assay (47). The purified G1-G2 was used for rotary shadowing electron microscopy (48), FACE analysis, and aggrecanase digests.
Keratan Sulfate Analysis by FACE Analysis-Purified G1-G2 containing 5 g of sulfated glycosaminoglycan was digested with glycosidases for analysis by FACE as described previously (49,50). Briefly, the digestion products recovered by centrifugation on Microcon YM-3 (3-kDa cutoff) spin columns (Millipore) were freeze-dried, resuspended in 5 l of 0.1 M 2-aminoacridone (dissolved in glacial acetic acid:Me 2 SO, 3:17 (v/v)), and allowed to stand for 15 min at room temperature. Thereafter, 5 l of 1 M sodium cyanoborohydride (freshly prepared in water) was added, and the samples were mixed and incubated at 37°C for 16 -24 h to complete the reductive amination reaction. The samples were cooled to room temperature, and 10 l of 37% (v/v) glycerol was added. Fluoro-tagged samples were stored at Ϫ70°C prior to electrophoresis on monogels. Equal amounts of glycosaminoglycan (0.5 g) were loaded on each gel. After electrophoresis, the gels were scanned immediately on a Storm PhosphorImager, and the bands were quantitated using Quantity One Software (Bio-Rad). The microstructure of keratan sulfate from recombinant and native pig G1-G2 was compared by measuring the ratio of monosulfated to disulfated disaccharides in each sample. The ratio of disaccharides in each lane is independent of the mass of sample loaded.
Digestion of G1-G2 with Aggrecanase-Conditioned medium was harvested from explants of pig articular cartilage (2.75 mg) cultured for 4 days in the presence of 1 M retinoic acid and 10 ng/ml interleukin-1␣ in serum-free DMEM. The medium was concentrated 9-fold on Millipore 10-kDa cutoff centrifugal filter units and used as a source of aggrecanase activity, as described previously (13). G1-G2 (1-3 g) was treated with or without glycosidases and then digested with 1 l of aggrecanase-containing medium at 37°C for the times given in buffer containing 50 mM Tris, pH 7.5, 0.1 M NaCl, and 10 mM CaCl 2 . The digestion was stopped by the addition of 10 mM EDTA and 2 mM 1,10-phenanthroline to inhibit aggrecanase activity. Equal amounts of G1-G2, with or without glycosidase or aggrecanase treatment, were loaded on polyacrylamide gels and analyzed by SDS-PAGE and Western blotting. The anti-NITEGE polyclonal neoepitope antibody (38) was used to detect the G1 fragment derived from aggrecanase cleavage.
Generation of an Anti-IGD Antibody-A rabbit polyclonal antibody raised against a synthetic peptide acetyl-CPDMELPLPRNITEG-amide was made by Quality Controlled Biochemicals Inc. The peptide sequence was chosen to elicit an antibody that would recognize the interglobular domain, but not the NITEGE neoepitope. The antibody reacted well with antigen containing the human sequence (above) by Western blotting and enzyme-linked immunosorbent assay (data not shown).

RESULTS
Recombinant human G1-G2 was expressed in primary bovine keratocytes, using a vaccinia virus expression system. To determine the viral load required for adequate G1-G2 expression, cells were co-infected with increasing amounts of vTF7-3 and vG1-G2 in the presence of [ 35 S]methionine (Fig. 1a). A single product of the expected size (approximate molecular mass, 120 kDa) was expressed by the co-infected cells (Fig. 1a, lanes 2, 4, 6, 8, and 10) but was absent in the control cells, which were infected with vTF7-3 alone (Fig. 1a, lanes 1, 3, 5, 7, and 9). Five plaque-forming units/cell generated readily detectable amounts of G1-G2, and scanning densitometry showed that the amount of secreted G1-G2 was proportional to the viral load, up to 20 plaque-forming units/cell (Fig. 1b). An inoculation of 5 plaque-forming units/cell vG1-G2 was used for all subsequent experiments. A time course experiment to determine the optimum infection time showed that G1-G2 was detected in the conditioned medium of the primary keratocytes 24 h post-infection, and its expression continued for up to 4 days (Fig. 1c). In contrast, COS-7 cells infected in the same way produced low levels of G1-G2 during the first 60 h but then died, presumably due to the burden of infection.
When [ 35 S]methionine-labeled G1-G2 was digested with keratanase, keratanase II, or E␤G, the size of the band was reduced by ϳ5 kDa (Fig. 1d, lanes 3-5). As expected, digestion with chondroitinase ABC did not change the size of 35 S-labeled G1-G2 (Fig. 1d, lane 2). The results indicate that the amount of keratan sulfate substituted on G1-G2 expressed in keratocytes (ϳ5 kDa) is substantially less than the amount (ϳ30 kDa) substituted on native G1-G2 from pig laryngeal aggrecan.
FACE Analysis of Keratan Sulfate on G1-G2-FACE analysis was done to confirm that the G1-G2 contained keratan sulfate and to determine the extent of sulfation. The specific Gal-GlcNAc6s and Gal6S-GlcNAc6S disaccharides were detected in keratanase II digests (Fig. 4a, lane 2). The ratio of the monosulfated:disulfated products, averaged from two separate preparations, was 1.00:0.90. The monosulfated disaccharide Gal-GlcNAc6S produced by keratanase II digestion was generated in equimolar yield to the equivalent monosulfated disaccharide GlcNAc6S-Gal produced by E␤G digestion. Faint bands migrating at the position of the unsulfated GlcNAc-Gal following E␤G digestion were barely visible on gels and were below the limit of quantitation. Similarly, sialic acid bands were faintly visible after digestion with neuraminidase, but they were too low to quantitate. No fucose was generated by fucosidase, suggesting that this modification was absent from G1-G2 (data not shown).
Unsulfated disaccharides present in E␤G digests were weakly visible (Fig. 4b, lane 6) but were below the limit for quantitation. E␤G digestion also generated a sialylated tetrasaccharide (Fig. 4b, lane 6). Neuraminidase treatment removed the capping sialic acid from this species and generated free sialic acid and the Gal-GlcNAc6S-Gal trisaccharide (Fig. 4b, lane 7). Overall, the results indicate that the keratan sulfate on G1-G2 expressed in bovine keratocytes is less sulfated than keratan sulfate chains on native pig G1-G2 and that the extent of sialic acid capping is low.
N-Linked Keratan Sulfate Influences Aggrecanase Cleavage-We have shown previously that almost all the keratan sulfate in the IGD is located on a 100-amino acid fragment of the IGD between the major and minor matrix metalloproteinase sites, at N 341 2 342 F and D 441 2 442 L, respectively (51,52). Barry et al. (29,31) have identified precise sites of keratan sulfate attachment within this region at Thr 352 , Thr 357 , Thr 370 and Asn 368 , clustered near the aggrecanase cleavage site. To determine whether G1-G2 expressed in bovine keratocytes carries both O-and N-linked keratan sulfate, we digested it with N-glycosidase F or O-glycosidase and monitored for loss of 5-D-4 epitope by Western blotting. Digestion with N-glycosidase F reduced 5-D-4 reactivity to faint bands (Fig. 5a, lanes 2  and 4) and decreased the size of G1-G2 by ϳ5 kDa (Fig. 5a, lane  4). O-Glycosidase treatment of G1-G2 did not decrease the size of G1-G2 or reduce immunoreactivity with 5-D-4 (data not shown). Collectively, these results strongly suggest that keratan sulfate on G1-G2 is N-linked and that threonine residues at positions 352, 357, and 370 are not substituted with keratan sulfate.
This interpretation is further supported by Western blotting of a duplicate gel with antibody 1-C-6 ( Fig. 5a, lanes 6 -8). The blot confirms that N-glycosidase reduced the size of G1-G2 by ϳ5 kDa (Fig. 5a, lane 7), whereas treatment with O-glycosidase did not reduce the size of G1-G2 (data not shown). Samples digested with a mixture of keratanases followed by N-glycosidase F (Fig. 5a, lane 8) were the same size as samples digested with N-glycosidase alone (Fig. 5a, lane 7), reiterating further that the keratan sulfate on G1-G2 is N-linked. It was interesting that a small amount of 5-D-4 immunoreactivity resisted N-glycosidase treatment in some cases (Fig. 5a, lane 4), but not others (Fig. 5a, lane 2). This is likely to reflect the fact that not only is 5-D-4 a particularly good anti-keratan sulfate antibody, but also that the antigen is polyvalent, so tiny amounts of residual antigen are detected with 5-D-4.
Several studies have shown that aggrecanase cleavage is blocked following deglycosylation of aggrecan with chondroitinase ABC and keratanases (36,37). We therefore investigated whether N-linked keratan sulfate in the IGD is sufficient to potentiate aggrecanase cleavage. Digestion of G1-G2 with keratanase, keratanase II, and E␤G, prior to digestion with aggrecanase, significantly reduced aggrecanase cleavage and generation of the 368 NITEGE 373 neoepitope (Fig. 5b). These results were reproducible in 9 of 11 experiments and suggested either that G1-G2 without keratan sulfate is a poor substrate for aggrecanase, as reported previously, or, alternatively, that G1-G2 without keratan sulfate is a poor antigen for the anti-NITEGE antibody. To resolve this issue, we compared duplicate Western blots developed with anti-NITEGE or with an antibody raised to a different sequence in the IGD (Fig.  5c). For both antibodies, signal for the aggrecanase-derived G1-NITEGE fragment was significantly diminished in G1-G2 samples that had been pretreated with keratanases (Fig. 5c,  lanes 2 and 4), suggesting that the reduced epitope reflects decreased digestion by aggrecanase and not diminished antibody reactivity. Digestion with N-glycosidase F prior to aggrecanase digestion also inhibited cleavage in the IGD (Fig. 5d), consistent with the finding that keratan sulfate on recombinant G1-G2 is N-linked. These results indicate that N-linked keratan sulfate is sufficient to potentiate aggrecanase cleavage in the IGD. DISCUSSION The importance of glycosylation in aggrecanolysis is an emerging theme, initiated by studies showing that deglycosylation of aggrecan inhibits cleavage at aggrecanase sites and suggesting that glycosaminoglycans are required for aggrecanase activity (36,37). On the other hand, exogenous glycosaminoglycans and highly sulfated polymers inhibit aggrecanase activity (32)(33)(34)(35), possibly by interactions with thrombospondin type I motifs (37,53) or cysteine-rich and spacer domains (54) in ADAMTS enzymes. This anomaly suggests that there may be competition between endogenous and exogenous glycosaminoglycans for binding to ADAMTS enzymes and further implies a direct effect of glycosaminoglycans on aggrecanolysis. Previous studies with keratan sulfate-deficient G1-G2 bac suggest that keratan sulfate may also influence aggrecanolysis, but by indirect effects. The mechanism by which keratan sulfate directly or indirectly modulates metalloproteinase activity has not been described, nor have the details of the keratan sulfate microstructure that might be involved in enzyme modulation.
Previously, we have done fragmentation studies analyzing regions of the IGD containing 32 amino acids (Phe 342 ϪGlu 373 ), 100 amino acids (Phe 342 ϪAsp 441 ), and IGD with the full G2 domain (ϳ110 kDa) (51,52). Based on size analysis, before and after keratanase treatment, these studies indicate that almost all the keratan sulfate in the IGD is localized on the 32-mer fragment in the region 350 DITVQTVTWPDMELPLPRNI-TEGE 373 that contains the four keratan sulfate substitution sites (underlined) identified by Barry et al. (29). Pratta et al. (36) have shown that removal of keratan sulfate from purified bovine aggrecan blocks cleavage at E 373 2 374 A; here we show that removal of N-linked keratan sulfate, presumably from Asn 368 , is sufficient to abrogate aggrecanase cleavage. This was achieved by deglycosylation with keratanases or with N-glycosidase F. Although we find no evidence for O-linked keratan sulfate on G1-G2 in the present study, we cannot eliminate the possibility that O-linked keratan sulfate may also have a role in aggrecanase potentiation, and additional studies are needed to address this issue.
To our knowledge, there is no evidence that aggrecan from human articular cartilage is ever found without keratan sulfate. However, the structure of keratan sulfate on human aggrecan changes dramatically with age (55). In contrast with aggrecan from young articular cartilage (0 -9 years), aggrecan from adult articular cartilage (18 -85 years) is substantially more sulfated and more highly modified by fucosylation and sialylation. Maturing cartilage (9 -18 years) has intermediate and increasing levels of sulfation, fucosylation, and sialylation (55). It is interesting to speculate that poten- FIG. 4. FACE analysis of keratan sulfate substituted on native and recombinant G1-G2. Keratan sulfate present on (a) purified G1-G2 expressed in bovine keratocytes or (b) purified G1-G2 from native pig laryngeal aggrecan was analyzed by FACE following digestion with keratanase II (KII; lanes 2 and 5) or following digestion with E␤G (lanes 3, 4, 6, and 7) with or without subsequent digestion with neuraminidase (lanes 4  and 7). 5. Aggrecanase and glycosidase digestion of G1-G2. a, G1-G2 expressed in bovine keratocytes was digested with (N) or without (-) N-glycosidase F, and 62.5 ng of sample was analyzed by Western blotting with 5-D-4. In a separate experiment, G1-G2 was digested with N-glycosidase F (N; lanes 4 and 7) or digested first with a mixture containing keratanase, keratanase II, and E␤G (KЈse), followed by digestion with N-glycosidase F (K'se N). Duplicate gels containing 50 ng of protein (lanes 3-5) or 0.5 g of protein (lanes 6 -8) were analyzed by Western blotting with 5-D-4 or 1-C-6. b, G1-G2, either undigested (U) or digested with keratanase (K), keratanase II (KII), or E␤G, was incubated with concentrated conditioned medium containing aggrecanases for an additional 8 h. The samples containing 0.25 g of protein were analyzed by Western blotting with anti-NITEGE antibody (␣-EGE). Lane C is the conditioned medium alone. c, G1-G2 incubated with or without a mixture of keratanases (KЈse) was incubated with concentrated conditioned medium containing aggrecanase for an additional 6 h. Samples containing 1 g of protein were analyzed for the NITEGE neoepitope (␣-EGE) (lanes 1 and 2) or an epitope present in the IGD (␣-IGD) (lanes 3 and 4). d, G1-G2 (0.5 g) digested with or without N-glycosidase F was incubated with concentrated conditioned medium containing aggrecanases for 4 h and analyzed for the NITEGE neoepitope (␣-EGE). Lane C is the conditioned medium alone. tial interactions of ADAMTS thrombospondin type I motifs (37,53) or cysteine-rich and spacer domains (54) with keratan sulfate may increase with increasing levels of sulfation and chain modification.
It is important to note that the studies on age-dependent changes of keratan sulfate described above were done with keratan sulfate from the keratan-sulfate rich region of aggrecan. We have shown recently, using aggrecan extracted from skeletally mature pig articular cartilage, that keratan sulfate in the keratan-sulfate rich region is not at all similar to keratan sulfate isolated from the IGD (56). Additional studies are required to determine whether keratan sulfate in the IGD changes with age, and, if so, whether those changes correlate with susceptibility to joint disease.
This study is the first to report keratan sulfate substitution on a recombinant protein. The amount of keratan sulfate on pig laryngeal G1-G2 that can be removed by keratanase treatment is ϳ30 kDa. In contrast, only 5 kDa keratan sulfate was present on G1-G2. This may reflect synthesis of shorter chains or a selective loss of substitution at O-linked sites because keratocytes have not been reported to produce O-linked keratan sulfate.
Unlike most cells in culture, primary bovine keratocytes continue keratan sulfate synthesis in vitro, albeit at reduced levels and with less sulfation (57,58). The presence of monoand disulfated disaccharides on G1-G2 indicates that the cells have the enzymes required for polylactosaminolactan synthesis, as well as the sulfotransferases required for keratan sulfate biosynthesis. This finding is consistent with studies by Funderburgh et al. (59), who detected mono-and disulfated keratan sulfate on endogenous proteoglycans synthesized by cultured bovine keratocytes.
One feature of in vitro keratan sulfate biosynthesis appears to be down-regulation of GlcNAc6ST (60), the sulfotransferase responsible for the sulfation of GlcNAc residues. Because GlcNAc6ST activity occurs simultaneously with chain elongation (61,62), down-regulation of GlcNAc6ST results in shorter chains with unsulfated disaccharides. In the context of keratan sulfate proteoglycans made by keratocytes in vitro, the G1-G2 reported here is unusual in that it has negligible amounts of unsulfated disaccharides, suggesting that GlcNAc6ST activity is not limiting in this system. Sulfation of galactose by KS-Gal6ST (63) and/or C6ST (64) does not appear to be limiting either because ϳ40% of the disaccharides were sulfated on galactose moieties. Thus, deficiencies in sulfotransferase activities do not appear to account for the reduced level of keratan sulfate on G1-G2. Approximately 60% of the pig laryngeal keratan sulfate was disulfated, compared with 40% for the rG1-G2, which may reflect the phenotype of the keratocytes. Similarly, the minimal sialic acid capping and the lack of fucosylation may reflect the keratocyte phenotype because corneal keratan sulfate is not fucosylated, and it may be capped with structures other than sialic acid, such as ␤GalNAc or ␣Gal (16,65).
In summary, we have produced and characterized a recombinant G1-G2 fragment that is substituted with keratan sulfate. Compared with keratan sulfate in the IGD of pig laryngeal aggrecan, keratan sulfate in the IGD of recombinant G1-G2 is less sulfated, has minimal sialic acid capping, and is exclusively N-linked to the core protein. Removal of the N-linked keratan sulfate diminishes aggrecanase cleavage at the E 373 2 374 A bond. The data indicate that N-linked keratan sulfate in the aggrecan IGD, most likely at position Asn 368 , confers susceptibility to aggrecanases.