INTRODUCTION

The functions of proteoglycans are highly diversified, ranging from mechanical functions, essential for maintaining the structural integrity of connective tissue, to effects on dynamic processes such as cell adhesion and motility and also cell differentiation and morphogenesis. Proteoglycans are the major components of the extracellular matrix, and their core proteins are substituted with glycosaminoglycans 10-fold their protein mass (1).

All glycosaminoglycans, except hyaluronic acid, are secreted as components of proteoglycans, covalently linked to a core protein. Chondroitin sulfate, dermatan sulfate, heparin, and heparan sulfate chains are bound to the protein by a serine-linked xylose-galactose-galactose bridge (2). The initial, apparently rate-limiting step in the biosynthesis of glycosaminoglycans is the transfer of xylose from UDP-xylose to serine residues of the core protein catalyzed by UDP-D-xylose:proteoglycan core protein -D-xylosyltransferase (EC 2.4.2.26) (XT)¹ (3, 4). Obviously, only selected serine residues were recognized by xylosyltransferase. Available information on the structure of proteoglycan core proteins has already shown that glycosylated serine residues are usually followed by a glycine residue (5). Further investigations based on comparison of the amino acid sequences of three chondroitin sulfate attachment sites in different proteoglycans suggested a recognition sequence S-G-X-G (X = variable amino acid) with some N-terminal acidic amino acids (6).

Here we aligned the amino acid sequences of 51 known chondroitin sulfate attachment sites in 19 different proteins to identify the complete recognition signal for XT defined by the primary protein structure.

One of these proteins was bikunin (Fig. 1), the inhibitory component of inter--trypsin inhibitor (ITI) (7). Bikunin carries a single chondroitin sulfate chain, which binds to the heavy chains of ITI, so this glycosaminoglycan is essential for the structure of the inhibitor.

Schematic structure of bikunin showing the chondroitin sulfate attachment site and the reactive sites of the two Kunitz-type domains 1 and 2.

EXPERIMENTAL PROCEDURES

Materials

UDP-[¹⁴C]xylose (9.88 kBq/nmol) was purchased from DuPont (Bad Homburg, Germany), Immobilon-AV membrane from Millipore (Eschborn, Germany), and nitrocellulose discs (inner diameter, 25 mm) from Sartorius (Göttingen, Germany). The liquid scintillation counter LS 500TD and scintillation mixture were supplied by Beckman (Fullerton, CA).

The Chang Liver cell line was obtained from ICN (Meckenheim, Germany), RNA Insta-Pure^TM was from Eurogentec (Seraing, Belgium), the enzymes used for cDNA cloning were from Boehringer Mannheim (Mannheim, Germany), the vector pET 15b and the E. coli strain BL21(DE3) were from Novagen (Madison, WI), and Chelating Sepharose and Q-Sepharose were from Pharmacia (Uppsala, Sweden). Oligonucleotides were synthesized by Genosys (Cambridge, United Kingdom). Minimum essential medium, fetal calf serum, serum-free and protein-free hybridoma medium, Pronase E, collagenase XI, antibiotic/antimycotic solution, trypsin, chymotrypsin, granulocyte elastase, kallikrein, VLR-pNA, Suc-AAPF-pNA, and Suc-AAA-pNA were obtained from Sigma (Deisenhofen, Germany).

The peptides QEEEGSGGGQKK, GVEGSADFLK, VCRSGSGLVGK, and PLVSSGEDEPK were synthesized by Quality Controlled Biochemicals (Hopkinton, MA). All other peptides were purchased from Bachem Biochemica (Heidelberg, Germany). Silk fibroin and hydrofluoric acid- and phenylmethylsulfonyl fluoride-degraded proteoglycan from bovine nasal septum cartilage were prepared as described previously (8). All other chemicals in pro analysi quality were purchased from Merck (Darmstadt, Germany).

Methods

Varying amounts of the potential acceptors were incubated with partially purified and enriched XT solution from chondrocyte culture supernatant and UDP-[¹⁴C]xylose. The reaction mixture for the assay contained, in a total volume of 100 µl: 50 µl of XT solution, 25 mM 4-morpholineethanesulfonic acid (pH 6.5), 25 mM KCl, 5 mM KF, 5 mM MgCl₂, 5 mM MnCl₂, 1.0 µM UDP-[¹⁴C]xylose, and varying acceptor concentrations. After incubation for 1 h at 34 °C, the mixtures were placed on small discs of Immobilon-AV membrane, which immobilizes even small peptides quantitatively by covalent links (9). After drying, the membrane discs were washed four times for 10 min with 0.1% Tween 20 in phosphate-buffered saline and measured by liquid scintillation counting. The enzyme activity was expressed in units (1 unit = 1 µmol of incorporated xylose·min¹) (10).

For a quantification of the acceptor activities of different proteins and peptides, the relation of the maximal rate (V_max) and the Michaelis-Menten constants (K_m) was determined for their xylosylation. K_m is inversely proportional to the affinity of XT to the acceptor, and V_max reflects the turnover number of the enzyme; thus, V_max/K_m can be defined as acceptor activity. For calculation of K_m and V_max, the transfer rates were measured as a function of the acceptor/substrate concentrations.

The reaction mixture contained, in a total volume of 100 µl: 50 µl of XT solution, 25 mM 4-morpholineethanesulfonic acid (pH 6.5), 25 mM KCl, 5 mM KF, 5 mM MgCl₂, 5 mM MnCl₂, 1.0 µM UDP-[¹⁴C]xylose, and varying acceptor concentrations. After incubation for 1 h at 34 °C, the mixtures were placed on small discs of Immobilon-AV membrane. After drying, the membrane discs were washed four times for 10 min with 0,1% Tween 20 in phosphate-buffered saline and measured by liquid scintillation counting.

The acceptor concentrations were calculated per xylosylation site. The potential xylosylation sites of silk fibroin are located in the repetitive hexapeptide GSGAGA. Silk fibroin consists of about 60% of this repetition (11), so the weight per mole of xylosylation sites is about 667 g/mol. Cartilage proteoglycan core protein with approximately 210 kDa has about 100 xylosylation sites (12-15) (2,100 g/mol). The recombinant bikunin with a short N-terminal leader sequence has a molecular mass of 17.5 kDa and contains one xylosylation site per molecule (7) (17,520 g/mol).

Small pieces of sternal cartilage were obtained during open heart surgery. The cartilage was cut aseptically, incubated for 90 min at 37 °C in 1% Pronase E in minimum essential medium and for several hours in 0.25% collagenase XI, until the tissue was digested (16). The free cells were grown in minimum essential medium supplemented with 10% fetal calf serum and antibiotic/antimycotic solution. After the cells reached confluence, they were incubated in serum-free medium for 4 days. The spent medium was harvested, and its proteins were separated by ion exchange column chromatography on Q-Sepharose. The proteins were bound to the resin in 50 mM Tris/HCl buffer (pH 8.0) and eluted by a gradient from 0 to 1.0 M NaCl in Tris/HCl buffer (pH 8.0).

cDNA cloning was performed using standard methods (17). RNA from Chang Liver cells was isolated using RNA Insta-Pure^TM. Bikunin cDNA was synthesized by reverse transcription with oligo(dT)₁₅ primer and Moloney murine leukemia virus reverse transcriptase and subsequent PCR with the bikunin specific primers 5-TCTCAGCATATGGCTGTGCTACCCCAAGAA (bikunin-N) and 5-GGCCAGGGATCCTCAGGAGAAGCGCAGCAG (bikunin-C) including NdeI or BamHI restriction sites at the 5-ends.

Synthesis of Recombinant [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²Bikunin

The cDNA coding for [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²bikunin² was synthesized by site-directed mutagenesis (Fig. 2). Parts of the bikunin cDNA were amplified in two PCRs with the primer pairs bikunin-N/5-TGGTCACTCCCACGCAGGGA and 5-CCTGCGGAGCCATCATCCA/bikunin-C. Each inner primer contained mutations for the two amino acid changes. In a third PCR, the complete bikunin cDNA including the mutations was amplified using the products of the two previous PCRs as "megaprimers." The amplified DNA was cut with the restriction enzymes BamHI and NdeI, ligated into the vector pET 15b, transformed, and expressed in the E. coli strain BL21(DE3) (18). The recombinant proteins carried a leader sequence with six histidine residues and were purified in one step by affinity chromatography with Ni²⁺-chelating resin.

Schematic presentation of cDNA synthesis of [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]¹bikunin using site-directed mutagenesis.

Characterization of the Inhibitory Activity of Recombinant Wild-type Bikunin and Recombinant [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²Bikunin

We tested the inhibitory activity of bikunin and the mutated bikunin variant against different serine proteinases. Different amounts of the inhibitors were added to the proteinases in a total volume of 150 µl. After a 30-min incubation at 37 °C, 50 µl of chromogenic substrate solutions with different concentrations were added. The optical density at 405 nm was measured periodically. The increase during the approximate linear phase was determined as a function of the substrate concentrations and served for calculation of the Michaelis-Menten constants. The K_m with and without inhibitor were used for calculation of the dissociation constants of the enzyme-inhibitor complex (K_i).

The end concentrations in the reaction mix were: 25 nM trypsin, 50 mM Tris/HCl (pH 7.8), 20 mM CaCl₂, 0.02% polyethylene glycol 6000, 0-1.2 µM bikunin or [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²bikunin, respectively, and 10, 50, 100, or 250 µM V-L-R-pNA.

The end concentrations in the reaction mix were: 25 nM chymotrypsin, 50 mM Tris/HCl, pH 7.8, 20 mM CaCl₂, 0.05% Triton X-100, 0-1.2 µM bikunin or [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²bikunin, respectively, and 10, 50, 100, or 250 µM Suc-A-A-P-F-pNA.

The end concentrations in the reaction mix were: 25 nM cathepsin G, 50 mM Tris/HCl, pH 7.8, 0-1.2 µM bikunin or [Val³⁶,Val³⁸]¹ [Gly⁹²,Ile⁹⁴]²bikunin, respectively, and 10, 50, 100, or 250 µM Suc-A-A-P-F-pNA.

The end concentrations in the reaction mix were: 25 nM granulocyte elastase, 50 mM Tris/HCl, pH 7.8, 0-1.2 µM bikunin or [Val³⁶,Val³⁸]¹, [Gly⁹²,Ile⁹⁴]²bikunin, respectively, and 10, 50, 100, or 250 µM Suc-A-A-A-pNA.

RESULTS

Alignment of the amino acid sequences of 51 known chondroitin sulfate attachment sites in unrelated proteoglycans (Table I) showed a significant homology and revealed a consensus sequence of 10 amino acids: a-a-a-a-G-S-G-a-b-a, where a = E or D and b = G, E, or D. 

Table I. Partial amino acid sequences of chondroitin sulfate attachment sites in different proteins The arrow over sequence indicates the glycosylated serine residue. Amino acids with a supposed importance for recognition by xylosyltransferase are boldface. CSF, colony-stimulating factor; PG, proteoglycan; MHC, major histocompatibility complex; a, acidic amino acid (E or D). Protein Chondroitin sulfate attachment site   Aggrecan (chondrocytes) (21, 22)  857E E S G A P D V S G D F T G S G  863D V S G D F T G S G D V S G H L  867D F T G S G D V S G H L D F S G  873D V S G H L D F S G Q L S G D R  892L P S G D L D S S G L T S T V G  910L T V E S G L P S G D E E R I E     A A P G V E D I S G L P S G E V     V E D I S G L P S G E V L E T A (part of a 28-fold repetition) Collagen IX (chondrocytes) (23)   40G H I Q G V E G S A D F L C P T CS-PG II (chondrocytes) (24) 1D  E A S G I G P E V P Versican (fibroblasts) (25)  499P E T S E H H F S G G E P D V F  517V P F H E E F E S G T A K K G A  553V S L F P E E S S G E I A I D Q  636T P R Q V V E L S G S S S I P I  647S S I P I T E G S G E A E E D E  842A S V F M E Q G S G E A A A D P  940G F P L E E D F S G D F R E Y S 1015E F T S S A E G S G E Q L V T V 1259E S D T E L L F S G L G S G E E 1331V Q T D I F E G S G S V T S T T 1434P S D L Y Y E P S G E G S G E E 1438Y Y E P S G E G S G E E V L P T 1772P S F Q P E F S S G A E E A L V Betaglycan (fibroblasts) (26)  408L V S D D L A N S G Q V G I P S  538R C A H L D R L S G R S C P A P Biglycan (fibroblasts) (27)   34L M M N D E E A S G A D T S G V   43E E A S G A D T S G V L D L D S  220M G G N P L E N S G F E P G A F Appican (fibroblasts) (28)  611A R A T E N E G S G L T N T P Q Fibronectin receptor (fibroblasts) (29)  259Y S V A V G E F S G D D T E D F  680T A P P E A E Y S G L V R H P G Vitronectin (fibroblasts) (30)   34D V D S P A E Y S G P E G S Y F  263G D G I D D F V S G V P R A A R PG 19 (fibroblasts) (31)    4F F P I S D D Y S G S G S G S G Decorin (fibroblasts) (32)    8D F M L E D E A S G I G P E V P Syndecan (epithelial cells) (33, 34)   29L P P E D Q D G S G D D S D N F   37S G D D S D N F S G A G A L Q D  198S Q L P A A E G S G E Q D F T F CD 44 (epithelial cells) (35)  172N P T D D D V S S G S S S E R S 346N S N V N R S L S G D Q D T F H Tenascin (brain) (36)  256Q C I C D S E Y S G D D C S E L  313E L R C P G D C S G K G R C A M Neurocan (brain) (37)  364R G D S E I P S S G D E G E I V  402D F Q E P L V S S G E D E P L D  826V P S T V D A T S G S E P K S S  936E G F W E E V A S G Q E D P T D CSF-1 (trophoblasts) (38)  301T N W V P E E A S G E A S E I P Platelet glycoprotein IIIa (39)  692R F Q Y Y E D S S G K S I L Y V Human MHC class II (40)  194S L E L E D P S S G L G V T K Q Apolipoprotein J (hepatocytes) (41)   90K F Y A R V C R S G S G L V G R  359H T S D S D V P S G V T E V V V Bikunin (hepatocytes) (7)    2V L P Q E E E G S G G G Q L V T Relative occurrence of D/E (%) 21 10 15 23 38 42 60 2 0 0 38 27 23 19 17 14 Relative occurrence of G (%) 6 8 4 12 4 12 0 23 0 98 4 19 6 15 6 10 Position  8  7  6  5  4  3  2  1 0 +1 +2 +3 +4 +5 +6 +7 Consensus sequence - - - a a a a G S G a a/G a - - -

The serum-free cell culture medium conditioned by human chondrocytes had an XT activity of about 0.8 milliunit/liter and a total protein concentration of 0.09 g/liter. Fractions from ion exchange chromatography eluted by a NaCl gradient from 0.40 to 0.45 M NaCl showed high XT activities. These fractions were collected and desalted. The XT activity of the preparation was about 4.5 milliunits/liter; the total protein concentration was 0.03 g/liter. So the specific XT activity referring to the total protein content was increased 17-fold from 8.9 milliunits/g to 152 milliunits/g.

K_m and V_max were determined for the xylosylation of different proteins and peptides (Table II). The relation V_max/K_m was defined as acceptor activity. As specific consensus sequence components appeared in proximity to the serine residue, the lower was K_m and the higher was the acceptor activity. Testing the tripeptide SGG and the peptide LNFSTGW, which contain a serine but do not match the consensus sequence, no xylose incorporation was detectable.

Table II. Michaelis-Menten (KM) constants and acceptor activities (Vmax/KM) for the xylosylation of different potential acceptor substrates by xylosyltransferase TFMS, trifluoromethanesulfonic acid; HF, hydrofluoric acid. Acceptor KM Vmax/KM µM Proteins   [Val36, Val38]1,[Gly92, Ile94]2bikunin 0.6 5133   Wild-type bikunin 0.9 3467   TFMS-degraded proteoglycan 155 15   HF-degraded proteoglycan 188 14   Silk (GSGAGA)n 545 5.5 Peptides     QEEEGSGGGQK (bikunin) 22 129   NFDEIDRSGFGFN 128 21      PLVSSGEDEPK (neurocan) 392 7.2    WAGGDASGE 418 6.1     VCRSGSGLVGK (apolipoprotein J) 943 2.5      GVEGSADFLK (collagen IX) 2,668 1.1      KTKGSGFFVF 3,625 0.7       KKDSGPY 8,273 0.3          SGG >50,000      LNFSTGW >50,000

In the reactive site of the two Kunitz-type domains of bikunin the amino acid sequence Met³⁶-Gly³⁷-Met³⁸ in domain 1 was changed to Val³⁶-Gly³⁷-Val³⁸ and in domain 2 Arg⁹²-Ala⁹³-Phe⁹⁴ was changed to Gly⁹²-Ala⁹³-Ile⁹⁴ by site-directed mutagenesis. The calculated K_i values for wild-type bikunin and the variant [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²bikunin with trypsin, chymotrypsin, cathepsin G, and elastase showed that the mutated bikunin variant lacked the inhibitory activity (Table III).

Table III. Dissociation constants (Ki) for the enzyme-inhibitor complexes of different serine proteinases and recombinant wild-type bikunin and the recombinant variant respectively Values for native ITI are also given. Proteinase Ki [Val36, Val38]1 [Gly92, Ile94]2 bikunin Wild-type bikunin ITI (29) M Trypsin (bovine) >104 2.3 · 109 2-10 · 1010 Chymotrypsin (bovine) >104 6.2 · 109 1-10 · 109 Cathepsin G (human) 4.2 · 105 4.5 · 107 2-6000 · 109 Elastase (human leucocytes) >104 2.8 · 108 6-15 · 108

DISCUSSION

We determined K_m and V_max for the xylosylation of different potential acceptors reflecting the affinity of XT to the acceptor and the turnover number, respectively (Table II). The acceptor activities (V_max/K_m) of the peptides confirmed the consensus sequence a-a-a-a-G-S-G-a-b-a revealed by comparison of chondroitin sulfate attachment sites in different proteins.

The acceptor tests showed that the minimal requirement for xylosylation was a polypeptide including serine and a C-terminally following glycine. For collagen IX alone, a chondroitin sulfate attachment site was described with alanine instead of glycine and an extent of glycosylation of about 70% (23). However, the sequence homologous peptide GVEGSADFLK showed only a weak acceptor activity in our test.

The acceptor activity of peptides including serine-glycine was higher, when more acidic amino acids were in the vicinity of serine, except for the positions directly adjacent to serine. In our acceptor test, acidic amino acids in the positions 5 to 2 and +2 to +4 as well as glycine residues in 1 and +3 improved the xylosylation rate by XT. The acidic amino acids were more effective, the closer their positions were to the serine.

In contrast to UDP-N-acetylglucosamine:polypeptide Nacetylglucosaminyltransferase, which transfers N-acetylglucosamine also to small peptides like NGT (42), XT recognizes only longer polypeptides. In our XT acceptor test, no xylosylation of the peptide SGG was observed and the acceptor activity of bikunin was 27-fold higher than that of the sequence homologous peptide with 11 amino acids.

In some core proteins, potential Ser-Gly xylosylation sites often escape xylosylation (43). We compared different proteoglycans and found that those with chondroitin sulfate chains of substantial importance for the function of the protein, e.g. human bikunin, match the consensus sequence better than proteoglycans that are naturally not quantitatively glycosylated, e.g. collagen and aggrecan.

Bikunin, the inhibitory component of the ITI, is quantitatively glycosylated by a chondroitin sulfate chain in position 10. This chain binds to the heavy chain of ITI and is essential for the structure of the inhibitor. The amino acid sequence of the chondroitin sulfate attachment site (⁵QEEEGSGGGQ¹⁵) corresponds to the determined consensus sequence.

An additional example for quantitatively glycosylated chondroitin-4 sulfate proteoglycan is the human C1q inhibitor (C1qI) (44). The amino acid sequence of this 30-kDa circulating complement inhibitor is still unknown, but it should be expected that the amino acid sequence of the chondroitin sulfate attachment site also matches the consensus sequence.

Among the tested acceptors, recombinant [Val³⁶,Val³⁸]¹, [Gly⁹²,Ile⁹⁴]²bikunin proved to be the most effective. The mutations in the reactive sites were far from the xylosylation site and should have no substantial effect on secondary protein structure. In wild-type bikunin, the N-terminal Kunitz-type domain 1 with the reactive site Met³⁶-Gly³⁷-Met³⁸ (position P₁-P₁-P₂) (45) is a competitive inhibitor for elastase and cathepsin G; domain 2 with Arg⁹²-Ala⁹³-Phe⁹⁴ inhibits proteinases like trypsin and chymotrypsin. The determined dissociation constants (K_i) for the proteinase/inhibitor complexes showed that the mutated bikunin variant had no inhibitory activity. Thus, in the XT assay, it should not interact with proteinases contained in the partially purified enzyme.

The acceptor activity of [Val³⁶,Val³⁸]¹,[Gly⁹²,Ile⁹⁴]²bikunin was even 1.6-fold higher than wild-type bikunin, and that of bikunin was 24-fold higher than of the synthetic peptide QEEEGSGGGQK. These results imply that the amino acid sequence of the recognition site is not the only regulatory factor that determines the priority for glycosylation of this site. Protein conformation is another important factor.