Limited proteolysis of human alpha2-HS glycoprotein/fetuin. Evidence that a chymotryptic activity can release the connecting peptide.

From the ‡Department for Clinical Chemistry and Clinical Biochemistry, The University of Munich, Nussbaumstrasse 20, 80336 Munich, Germany, the §Max-Planck Institute for Biochemistry, Am Klopferspitz, 82152 Martinsried, Germany, the ¶Behringwerke AG, P. O. Box 1140, D-35007 Marburg, Germany, and the iInstitute for Physiological Chemistry and Pathobiochemistry, The University of Mainz, Duesbergweg 6, D-55099 Mainz, Germany

␣ 2 -HS glycoprotein is a major protein of human plasma whose function is still obscure. A proteolytically processed form of ␣ 2 -HS glycoprotein lacking a segment of 40 amino acid residues bridging its heavy and light chain portions ("connecting peptide") has been described suggesting that this peptide is released by posttranslational processing to fulfill biological role(s) of ␣ 2 -HS glycoprotein. To test this hypothesis we investigated how the connecting peptide is released from the parental molecule by limited proteolysis. We developed monoclonal antibodies to various portions of the connecting peptide and its NH 2 -terminal flanking region which cross-react with the native ␣ 2 -HS glycoprotein. Purified ␣ 2 -HS glycoprotein from human plasma was subjected to limited proteolysis by proteinases including trypsin, chymotrypsin, elastase plasmin, kallikrein, thrombin, and renin. Immunoprint analysis of the proteolytic digests indicated that ␣ 2 -HS glycoprotein is readily cleaved in its connecting peptide region. NH 2terminal amino sequence analysis of the generated fragments demonstrated that a single proteinase, chymotrypsin, cleaves the critical Leu-Leu bond flanking the NH 2 -terminal portion of the connecting peptide region. Most but not all of the other proteinase cleavage sites map to a short stretch of 9 residues located in the center portion of the connecting peptide region. Immunoprint analysis of plasma samples from patients with sepsis demonstrate that the connecting peptide region is cleaved under pathological conditions. Our results indicate that the connecting peptide and/or fragments thereof are readily releasable from ␣ 2 -HS glycoprotein in vitro and in vivo.
Fetuins occur in large amounts in blood and cerebrospinal fluid and accumulate to high concentrations in calcified bone (4). The fetuins form a family within the superfamily of mammalian cystatins (5); they are characterized by a tripartite structure, i.e. two NH 2 -terminally located cystatin-like domains (D1, D2) and a unique COOH-terminal domain (D3) not present in other mammalian cystatins (6). Although many biological functions have been attributed to fetuins, none has been unequivocally established to date (for review, see Ref. 7).
The human homologue of fetuin, ␣ 2 -HS glycoprotein, has long been known as a major protein of the plasma and the cerebrospinal fluid. Early studies have demonstrated that ␣ 2 -HS isolated from Cohn's fraction VI of human plasma is a two-chain molecule consisting of a A-chain of 282 amino acid residues and a B-chain of 27 residues (8,9). Cloning of the ␣ 2 -HS cDNA revealed that a single mRNA transcript encodes the A and the B chains, and that a sequence segment of 40 residues ("connecting peptide") bridges the A and B portions in the primary translation product, pre-␣ 2 -HS (10). Because this segment is absent in the ␣ 2 -HS form of Cohn's fraction VI it has been hypothesized that the connecting peptide is released from the precursor protein by an unknown proteinase, and that the liberated peptide might serve a biological function (9). We isolated ␣ 2 -HS from freshly drawn human plasma in the presence of proteinase inhibitors, and determined that the circulating form of ␣ 2 -HS is a two-chain protein with a heavy chain of 321 residues and a light chain of 27 residues (11). This circulating form of ␣ 2 -HS still contained the connecting peptide attached to the heavy chain, and only the terminal residue of arginine in position 322 was missing. These findings prompted us to investigate ␣ 2 -HS proteolytic processing thereby addressing the following questions: (i) is the release of the connecting peptide a consequence of the isolation procedure or does it occur under (patho)physiological conditions; (ii) is the connecting peptide released in a single step, i.e. by cleavage of the Leu 282 -Leu 283 bond, or by sequential processes; (iii) which proteinase(s) is (are) involved in the release of the connecting peptide or fragments thereof, and what is the nature of the enzyme(s), if any, that split(s) the critical Leu-Leu bond? Our experimental data indicate that the connecting peptide region is highly succeptible to proteolytic breakdown in vitro and in vivo, and that a chymotryptic activity can release the connecting peptide or fragments thereof from the circulating form of ␣ 2 -HS. boxypeptidase Y (from yeast) from Boehringer Mannheim; and renin (from human kidney) from Medor, Gauting, Germany. Aprotinin (from bovine lung) was a gift from Bayer. Protein purity was assessed by polyacrylamide gel electrophoresis (PAGE) in the presence of sodium dodecyl sulfate (SDS), see below. Peptides (Table I) were synthesized by a modified Merrifield procedure (12) on an Applied Biosystems synthesizer, Model 430 A, and purified by reverse-phase high performance liquid chromatography (HPLC) on a large pore Vydac TP RP 10 column; their amino acid sequences were confirmed by Edman sequencing. Peptides were coupled to bovine serum albumin or keyhole limpet hemocyanin by carbodiimide. Antibodies were biotinylated by biotin-⑀aminocaproyl-N-hydroxysuccinimide (13), and linked to horseradish peroxidase by N-succinimidyl-3-(2-pyridyldithio)-propionate (14).
The proteolytically trimmed form of ␣ 2 -HS containing the A-chain was isolated from Cohn fraction VI of human plasma according to the original procedures (8,16). For the recombinant expression of ␣ 2 -HS domain 3, a corresponding cDNA fragment (positions 760-1490; 9) was generated by polymerase chain reaction, ligated into a baculovirus transfer vector (17), and expressed in Sf9 insect cells (18). Recombinant domain 3 covers positions 235-349 of the ␣ 2 -HS protein sequence and is preceded by a single Asp residue missing in the authentic sequence.
Production of Antibodies to ␣ 2 -HS-Monoclonal antibodies to native ␣ 2 -HS were produced by the standard technique (19) with minor modifications (20). Female BALB/c mice were immunized, and harvested spleen cells were fused with X63./Ag8.653 myeloma cells using 50% (w/v) polyethylene glycol 4000 (Merck) as the fusogenic agent. Hybridoma cells were selected and screened for the production of specific antibodies by the indirect enzyme-linked immunosorbent assay (ELISA) using titer plates coated with the two-chain form of ␣ 2 -HS of 348 residues, the synthetic light chain of 27 residues, or the synthetic connecting peptide of 40 residues. Colonies producing antibodies of the desired specificity were subcloned three times by limiting dilution. Ascites was obtained from mice with multiple myelomas induced by intraperitoneal injection of 2,6,10,14-tetramethyl-pentadecane (Pristane, Aldrich) followed after 10 days by an injection of 5 ϫ 10 6 hybridoma cells in RMPI 1640. Polyclonal antibodies against ␣ 2 -HS, albumin, and IgG purified from human plasma were produced in female New Zealand White rabbits following standard immunization protocols.
Purification of Antibodies-Monoclonal antibodies were precipitated from mouse ascites with a saturated (NH 4 ) 2 SO 4 solution followed by immunoaffinity chromatography on matrix-bound ␣ 2 -HS. Briefly, antibodies were precipitated at 45% (NH 4 ) 2 SO 4 saturation. Following centrifugation (4000 ϫ g, 30 min, 4°C) the sediment was washed with a (NH 4 ) 2 SO 4 solution of 50% saturation, and the precipitated antibodies were collected as above. The pellet was dissolved in PBS, pH 7.3, and dialyzed against the same buffer prior to immunoaffinity chromatography on an ␣ 2 -HS-Sepharose 4B column (10 mg of protein/1 g of gel). The purity of the antibody preparation was monitored by analytical SDS-PAGE and isoelectric focussing, pH 3-9, on precast gels using the phast system according to the instructions of the manufacturer (Pharmacia). Polyclonal antibodies from rabbit antisera were purified by immunoaffinity chromatography using the respective antigens.
Limited Proteolysis of ␣ 2 -HS Glycoprotein-Proteolysis of ␣ 2 -HS was carried out in PBS, pH 7.2, at a protein concentration of 1 mg/ml. To optimize the conditions of limited proteolysis, a first set of experiments was carried out on an analytical scale (10 g of ␣ 2 -HS per experiment) where ratios of enzyme over substrate ranged from 1:100 to 1:10,000 (w/w). After 1 h of incubation at 37°C the reaction was stopped by addition of an equal volume of SDS sample buffer and boiling for 5 min at 95°C, followed by analytical SDS electrophoresis. In a second set of experiments the proteolytic digestion was optimized with respect to time at a fixed enzyme over substrate ratio. The experiments were carried out as above except that 100 g of ␣ 2 -HS (total amount) was incubated with the optimized concentration of the proteinase in a final volume of 100 l, and aliquots of 10 l were withdrawn at fixed time intervals. The resultant protein mixtures were subjected to SDS-PAGE under reducing conditions and visualized by silver staining. For largescale preparation of cleavage products, 50 mg of ␣ 2 -HS was cleaved at 37°C under optimized conditions. The proteolytic reactions were stopped by the addition of benzamidine-HCl (final concentration 1 mM) and stored at Ϫ20°C until analyzed.
Isolation and Characterization of ␣ 2 -HS Fragments-Fifty mg of ␣ 2 -HS was subjected to limited proteolysis with chymotrypsin, elastase, kallikrein, plasmin, or thrombin, and the cleavage mixtures were fractionated by gel filtration on Superose 12 fast flow (Pharmacia) in a buffer containing 0.05 M Na 2 HPO 4 , 0.15 M NaCl, pH 7.2, at a flow rate of 0.2 ml/min. For each run 4 mg of protein was applied. Protein elution was monitored at 280 nm and fractions were pooled and rechromatographed by reverse-phase HPLC on a Hibar-Lichrospher 125-4 column (Merck) using a linear gradient of 0 -70% (v/v) acetonitrile formed by mixing 0.1% (v/v) trichloroacetic acid in H 2 O and 0.1% trichloroacetic acid in acetonitrile as detailed earlier (11).
Amino Acid Sequence Analysis-Peptides isolated from the proteolytic digests of ␣ 2 -HS were sequenced employing Edman chemistry (25) on a gas-phase sequencer 477A (Applied Biosystems) running standard software. The corresponding phenylthiohydantoin derivatives were analyzed by an on-line HPLC gradient system 120A (Applied Biosystems). Amino-terminal amino acids were determined manually following the method of Chang et al. (26).
Nomenclature of Antibodies, Peptides, and Proteins-MAHS1 to

Isolation and Characterization of ␣ 2 -HS from Human
Plasma-␣ 2 -HS was isolated from freshly drawn human plasma by two succesive immunoaffinity steps: capture of ␣ 2 -HS on anti-␣ 2 -HS-Sepharose, and removal of remaining contaminating proteins on anti-albumin-Sepharose and anti-immunoglobulin-Sepharose, respectively. To prevent proteolytic modification during the isolation procedure, proteinase inhibitors were added to the chromatography buffers. The characterization of the isolated protein by SDS-PAGE and immunoprinting showed a single band of 58 kDa reacting with polyclonal antibodies directed to ␣ 2 -HS (not shown). The NH 2 -terminal amino acid sequence analysis revealed two sequences in equimolar amounts, Ala-Pro-His (heavy chain) and Thr-Val-Val (light chain). The COOH-terminal amino acids were determined as threonine (heavy chain) and valine (light chain). This form of ␣ 2 -HS containing the heavy chain (residues 1-321) and light chain (323-349) was used throughout the study unless otherwise indicated.
Production of Monoclonal Antibodies to ␣ 2 -HS-Mice were immunized with purified ␣ 2 -HS (see above), their spleen cells were removed and fused with myeloma cells to obtain hybridomas secreting antibodies to ␣ 2 -HS. Clones reacting positive in ELISA with intact ␣ 2 -HS were further subtyped by immunoprint analysis for specific recognition of the heavy chain, the connecting peptide or the light chain (Fig. 1). Antibodies of 5 clones (MAHS1, MAHS2, MAHS3, MAHS4, and MAHS7) recognized the ␣ 2 -HS heavy chain (53 kDa, lane 2) but failed to react with the light chain (5 kDa, lane 2). Antibodies MAHS3 and MAHS7 but not MAHS1, MAHS2, and MAHS4 reacted with the connecting peptide (4 kDa, lane 3), suggesting that the latter group of antibodies recognize the A-chain portion of ␣ 2 -HS. The MAHS antibodies cross-reacted well with plasma ␣ 2 -HS except for MAHS4 which weakly recognized purified ␣ 2 -HS but failed to detect ␣ 2 -HS in plasma (lanes 1 and 2). . Furthermore, antibody MAHS1 decorated a degradation product of D3 that is not stained by antibodies MAHS2 or MAHS4. For comparison, antibody MAHS7 directed to the connecting peptide, and a polyclonal antibody raised against native ␣ 2 -HS are included (Fig. 2, bottom panels). Because antibodies MAHS1, MAHS2, and MAHS4 recognize domain D3 (positions 235-349) but not the connecting peptide (283-322) or the light chain (323-349) they are likely to be directed to the COOH-terminal portion of 48 residues of the A-chain (235-282).
Epitope Mapping of MAHS Antibodies-To discriminate between epitopes recognized by the various MAHS antibodies a competitive ELISA was performed. In this assay the binding of a biotinylated antibody to its antigen is measured in the presence of increasing concentrations of unlabeled competitor antibodies; displacement of the biotinylated antibody by an unlabeled antibody indicates that the two antibodies recognize closely neighbored or even overlapping epitopes. The unlabeled cognate antibody and an unrelated antibody were used as the positive and negative controls, respectively. Of the antibodies to the A-chain, MAHS1 and MAHS2 bound to two distinct epitopes whereas MAHS4 competed with antibody MAHS2 for the same epitope (not shown). Antibodies to the connecting peptide, MAHS3 and MAHS7, did not displace each other and hence bind to two distinct epitopes (not shown). Hence the set  of MAHS antibodies covers at least 4 distinct epitopes exposed by the ␣ 2 -HS heavy chain.
For the precise epitope mapping in the connecting peptide region, we employed an indirect ELISA using the synthetic connecting peptide of 40 residues, and three peptides of 12-16 residues covering the NH 2 -terminal portion (LAA-12), the center part (HYD-12), and the COOH-terminal segment (SLG-16), respectively of the connecting peptide (Table I). As a control, antibody MAHS1 was applied which reacts with ␣ 2 -HS but fails to bind to the connecting peptide or fragments thereof (Fig. 3A). MAHS3 bound to ␣ 2 -HS, the connecting peptide, and peptide SLG-16, but not to peptides LAA-12 and HYD-16 (Fig. 3B) indicating that this antibody recognizes an epitope associated with the COOH-terminal portion (positions 307-322) of the connecting peptide region. MAHS7 recognized ␣ 2 -HS, the connecting peptide, and peptide HYD-12 but not peptides LAA-12 and SLG-16 (Fig. 3C), demonstrating that this antibody binds to an epitope in the center portion of the connecting peptide (295-306). None of the MAHS antibodies reacted with the NH 2 -terminal segment of the connecting peptide (233-294). Taken together our data indicate that we have produced 5 monoclonal antibodies to 4 distinct epitopes of ␣ 2 -HS which we arbitrarily name A (MAHS1), B (MAHS2, MAHS4), C (MAHS7), and D (MAHS3). Clustering of the epitopes on the COOH-terminal portion of the heavy chain suggests that this segment is immunodominant within the ␣ 2 -HS molecule.
Limited Proteolysis of ␣ 2 -HS-To establish a possible sequence of proteolytic events leading to the removal of the connecting peptide, we employed limited proteolysis and Western blot analysis with MAHS antibodies. In a first set of experiments, we optimized the conditions of limited proteolysis such that the formation of high molecular weight cleavage products is favored. A constant amount of ␣ 2 -HS was incubated with increasing concentrations of trypsin, chymotrypsin, elastase, plasmin, kallikrein, thrombin, or renin. The enzyme over substrate ratios varied from 1:50 to 1:10.000. In a second set of experiments, we varied the time of proteolysis from 5 min to 8 h, with enzyme:substrate ratios remaining constant. The digests were analyzed by SDS-PAGE and densitometric scanning of the silver stained gels. Fig. 4 illustrates a representative experiment where the proteolytic cleavage of ␣ 2 -HS by increasing amounts of trypsin was monitored for 1 h. At a 10,000-fold excess of the substrate, trypsin splits the 58-kDa parental molecule mainly into a fragment of 53 kDa (lanes 1  and 2). Increasing concentrations of trypsin (1:1000 to 1:50) caused the progressive breakdown of ␣ 2 -HS into fragments ranging from 18 to 35 kDa (lanes 3-5). Using a fixed ratio of 1:1,000 and incubation periods ranging from 5 min to 8 h we determined that the initial cleavage product of 53 kDa peaked only after 5 min of incubation (not shown). These conditions (enzyme over substrate ratio ϭ 1:1000, t ϭ 5 min) were selected for the limited proteolysis of ␣ 2 -HS by trypsin. Accordingly we optimized the conditions for the other enzymes (Table II). All selected proteinases readily cleaved ␣ 2 -HS except for renin which failed to process ␣ 2 -HS even at equimolar conditions (not shown). The optimum conditions varied among the different proteinases in that the enzyme over substrate ratios ranged from 1:100 (thrombin) to 1:10,000 (chymotrypsin), whereas the incubation periods varied from 5 min (trypsin at 1:1000) to 7 h (thrombin at 1:100). Under these conditions major cleavage products of 48 -53 kDa were generated from 50 to 90% of the initial ␣ 2 -HS (Fig. 5, lanes 3-8). In the case of elastase, an additional major degradation product of about 40 kDa was formed (lane 6). For comparison ␣ 2 -HS preparations containing the heavy chain (lane 1) and the A-chain (heavy chain without connecting peptide), respectively (lane 2), were included in the analysis.
Immunoprint Analysis of the Proteolytic Digests-To study   Table I. the cleavage of the connecting peptide region, we analyzed the digests for the presence of the connecting peptide by immunoprinting with antibodies MAHS1, MAHS2, MAHS3, or MAHS7 (Fig. 6). Antibodies MAHS1 and MAHS2 directed to the Achain portion of ␣ 2 -HS recognized the intact heavy chain of 58 kDa (upper panels, lane 1), the major cleavage products of 48 -53 kDa (lanes 2-6), and the minor fragments of 18 -35 kDa (e.g. lanes 4 and 5). In contrast, antibodies MAHS3 and MAHS7 directed to the COOH-terminal and the center part, respectively, of the connecting peptide region detected the intact heavy chain but none of the cleavage products. This finding indicates that the connecting peptide region is readily accessible and therefore a primary cleavage site for proteinases in the ␣ 2 -HS molecule. The failure of MAHS7 to detect any cleavage products of ␣ 2 -HS strongly suggests that the initial proteolytic cuts occur around the MAHS7 epitope, namely at positions 295-306 of the ␣ 2 -HS heavy chain. A time resolved analysis of the limited proteolysis indicated that the cleavage of the connecting peptide region in every case preceded the cleavage of the flanking segments (not shown).
Isolation of ␣ 2 -HS Fragments-To identify the proteolytic cleavage sites in ␣ 2 -HS the digests were size fractionated by preparative gel filtration. Fractions containing fragments of Յ20 kDa were pooled, concentrated, and subjected to reversephase HPLC (not shown). The separated peaks were collected manually and analyzed by quantitative amino acid analysis (not shown) and by manual NH 2 -terminal amino acid sequence analysis performing three cycles per peptide. Table III presents 17 distinct fragments identified in the proteolytic digests generated with plasmin, kallikrein, elastase, or chymotrypsin.
Most of the fragments (13 out of 17) are derived from domain D3 of ␣ 2 -HS. As shown in the graphic representation (Fig. 7, top  panel), 9 out of these 13 fragments originate from the connecting peptide region whereas 4 fragments map to the flanking regions, i.e. the proline-rich region (single fragment) and the light chain region (three fragments). The remaining fragments are produced by the cleavage of the NH 2 -terminal domain D1. These findings confirm that the connecting peptide region represents the major proteinase-sensitive site of ␣ 2 -HS.
In Vivo Processing of ␣ 2 -HS Probed by Monoclonal Antibodies-Does proteolytic processing of ␣ 2 -HS also occur in vivo? To answer this question we examined human plasma samples by SDS-PAGE and Western blotting applying MAHS antibodies. Fifteen plasma samples from healthy donors and from patients suffering from polytrauma and sepsis were probed with monoclonal antibodies directed to the connecting peptide. A repre-   Fig. 5), and electrotransferred to nitrocellulose; four identical replicas were made. Immunodetection of the fragments was performed with monoclonal antibodies MAHS1 (epitope A), MAHS2 (epitope B), MAHS3 (epitope D), and MAHS7 (epitope C). Bound antibodies were detected with biotinylated secondary antibody (1:2,000; anti-mouse immunoglobulin, from rabbit) and pre-formed biotin-avidin-peroxidase complex (2 g/ml), followed by 4-chloro-naphthol/H 2 O 2 . a Molar ratio of enzyme over substrate. b Molecular masses of the major cleavage products. c Yields of major cleavage products are judged from densitometric tracings of the gels (starting product ϭ 100%). sentative experiment including normal plasma (lane 1), purified ␣ 2 -HS (lane 2), and three pathological plasma samples (lanes 3-5) is presented in Fig. 8. Four identical replica blots were made, and total protein was visualized with colloidal gold staining (left panel). Immunoprinting with a polyclonal antibody to ␣ 2 -HS (right panel) indicates that similar amounts of ␣ 2 -HS and/or its degradation products have been applied to the various lanes (note that ␣ 2 -HS is a negative acute phase reactant, i.e. its plasma concentration decreases during inflammation). In the control plasma ␣ 2 -HS migrated as a single band of 58 kDa whereas two closely spaced bands of 53/55 kDa were separated in the pathological plasma samples; they most probably represent in vivo degradation products of ␣ 2 -HS. Immunoprinting by monoclonal antibodies MAHS3 or MAHS7 revealed that the ␣ 2 -HS protein present in normal plasma contained the connecting peptide region (center panels, lane 1). In contrast, ␣ 2 -HS from the pathological plasma samples was not decorated by either of these antibodies, suggesting that ␣ 2 -HS was cleaved in the connecting peptide region. Similar results were obtained with pathological plasma samples from 9 additional donors (not shown) suggesting that the release of the connecting peptide or fragments thereof does occur in vivo and is a common event in septicemia.

DISCUSSION
Fetuins are abundant in mammalian plasma and tissues. Much knowledge has accumulated on the structure of these prototypic glycoproteins, however, their functional role is less well defined. Bovine fetuin has been implicated in the stimulation of cell growth (27) and the modulation of brain development (28). The rat homologue, first described as phosphoprotein pp63, was shown to be an inhibitor of the endogenous tyrosine kinase activity associated with the insulin receptor (29,30). The fetuins have therefore been implicated in the regulation of the mitogenic response following insulin stimulation (29 -31). The fact that fetuins bind transforming growth factor-␤-like cytokines (32) and therefore can antagonize the action of transforming growth factor-␤ and bone morphogenic proteins similarly suggest a role for fetuins as modulators of growth and development.
Consistently, fetuins have been localized in mineralized bone tissue. Rat fetuin has also been described as bone sialic acid containing protein (33) or 60 K acidic glycoprotein (34), and both proteins were introduced as putative modulators of bone formation. A rabbit bone protein, hemonectin, which is closely associated or identical with rabbit fetuin (35) was originally defined as a homing factor for the granulocyte lineage (36). The human counterpart of fetuin, ␣ 2 -HS glycoprotein, is thought to play a major role in bone metabolism (37). To this end we have recently shown that domain D1 of fetuins mediates inhibition of apatite formation to inhibit unwanted mineralization in plasma and possibly during early stages of bone mineralization (18).
Although several functions have now been proposed for mammalian fetuins our knowledge about molecular mechanisms regulating these activities is still incomplete. Serine phosphorylation was shown to be critically required for rat fetuin inhibition of the insulin receptor tyrosine kinase (29) and multiple serine phosphorylation was also demonstrated in human fetuin (38). In contrast, the inhibition of apatite formation by fetuins was largely independent of secondary amino acid modifications (18). One intriguing structural feature of human fetuin with possible functional implication was revealed when the cDNA sequence of human ␣ 2 -HS was published (10) and compared to the known protein sequences from the A-and B-chains (8,9). Namely, a stretch of 40 amino acids predicted by the cDNA sequence was conspicuously absent from the published protein sequence. This tempted us and  for chymotrypsin (C 1 to C 4 ), elastase (E 1 to E 3 ), kallikrein (K 1 to K 6 ), and plasmin (P 1 to P 4 ) are given. MAHS epitopes A and B are hatched. Lower panel, the amino acid sequence of the connecting peptide including its flanking residues: epitopes C and D of MAHS7 and MAHS3, respectively, are marked by thin lines, and protease cleavage sites are indicated by arrows. The processing sites for the unknown endoproteinase (En) generating the circulating two-chain form of ␣ 2 -HS, and for the exopeptidase (Ex) releasing the terminal residue of Arg 322 are also given.  5) were separated by SDS-PAGE (10 -20%) under reducing conditions followed by electroblotting onto a nitrocellulose membrane; 4 identical replicas were made. Left panel, total protein staining with colloidal gold. The center panels are immunoprints with monoclonal antibodies MAHS3 (epitope D) and MAHS7 (epitope C), and the right panel represents an immunoprint with polyclonal antiserum AS5359 to ␣ 2 -HS. Bound antibodies were detected by incubation with 1 g/ml peroxidase-conjugated secondary antibody (anti-mouse or anti-rabbit immunoglobulin), followed by 4-chloronaphthol/H 2 0 2 . others to speculate that this connecting peptide served some unknown biological function after being proteolytically liberated. Subsequently we have isolated from freshly drawn plasma under the strict protection of proteinase inhibitors a two-chain form of ␣ 2 -HS where the heavy chain still harbors the connecting peptide except for a single COOH-terminal residue of Arg at position 322 (11). This heavy chain (as opposed to A-chain) form of ␣ 2 -HS is dominant in the plasma; we have failed to detect in plasma a single chain form or the two-chain form with the heavy chain including Arg 322 (11). We have, however, identified the single chain form along with the heavy chain form of ␣ 2 -HS in the supernatant of a human hepatoma cell line, HepG2, suggesting that the endoproteolytic cleavage and the exoproteolytic removal of Arg 322 are closely associated with the site of biosynthesis (38) and most probably do not occur in the plasma. Still an explanation was missing as to why a rather homogenous A-chain form of ␣ 2 -HS has been isolated by alternative purification procedures (8,9,16). Predictably, a single clip at positions 282/283 of the heavy chain of ␣ 2 -HS should release a connecting peptide of 39 residues (desArg 40 -CP) and generate the A-chain form.
To search for proteinases that could release the connecting peptide in vitro, we subjected our plasma ␣ 2 -HS to limited proteolysis by a set of proteinases and analyzed the fraction of small peptides liberated in this way. A single enzyme, chymotrypsin, cleaves the critical Leu-Leu bond at positions 282/283 implying that a chymotrypsin-like activity might release de-sArg 40 -CP in vivo. An additional cleavage site of chymotrypsin at positions 283/284 (Leu-Ala) is adjacent to the former site and might yield a shortened form of the connecting peptide, desLeu 1 desArg 40 -CP. Given the fact that the molar enzyme over substrate ratio was 1:10,000, and that Ͼ90% of the starting material was processed after an incubation period of 30 min, it appears that chymotrypsin is an extremely efficient processing enzyme for ␣ 2 -HS. Because two additional chymotrypsin cleavage sites are located distally of the critical Leu-Leu bond we cannot rule out the possibilities that the connecting peptide is released sequentially by stepwise cleavage from its COOH-terminal end, or that the released connecting peptide is subsequently cleaved into fragments of variable size.
Unlike chymotrypsin most of the other tested proteases split the center portion of the connecting peptide. Six fragments are released by the cleavage of a short segment of 9 residues harboring the MAHS7 epitope (positions 294 to 303; Ala-His-Tyr-Asp-Leu-Arg-His-Thr-Phe), Fig. 7, bottom panel. This part of the molecule is well conserved in the fetuin family: 5 out of 9 amino acids are invariant among human, bovine, sheep, and rat fetuins (6). In contrast, the NH 2 -terminal portion of the connecting peptide that harbors 2 cleavage sites, and its COOH-terminal portion where we found no cleavage site are less conserved (2/12 and 8/20, respectively). Clustering of the cleavage sites on a short segment indicates that this stretch is likely to represent a primary attack site for proteinases in ␣ 2 -HS akin the "hyperfrangible region" identified in human high molecular weight kininogen (39). This conclusion is supported by the finding that the MAHS7 epitope is readily lost even under strictly limited proteolysis in vitro.
Is the connecting peptide released in vivo? We have been unable to find in anticoagulated plasma of healthy donors significant amounts of the A-chain form which is devoid of the connecting peptide (not shown). However, under pathological conditions, such as sepsis, truncated forms of ␣ 2 -HS were detected in the plasma, and these forms were virtually free of the connecting peptide. These observations are compatible with the notion that the connecting peptide (or at least major part of it) is released from the circulating two-chain form under patho-logical conditions. We cannot, however, rule out that ␣ 2 -HS, sequestered from the plasma compartment or locally synthesized, might release its connecting peptide in situ due to the action of chymotryptic activities. Future studies will focus on the proteolytic processing of ␣ 2 -HS in peripheral organs such as the bone, and will directly address the potential biological roles of the connecting peptide and its major fragments.